Files that changed from the base of the PR and between 5ab305c and c37b715.

• sleap/config/pipeline_form.yaml (2 hunks)
• sleap/gui/learning/dialog.py (1 hunks)
• sleap/gui/widgets/video.py (2 hunks)
• sleap/nn/inference.py (12 hunks)
• sleap/nn/tracking.py (15 hunks)
• sleap/util.py (2 hunks)
• tests/nn/test_tracker_components.py (2 hunks)
• tests/nn/test_tracking_integration.py (4 hunks)

tests/nn/test_tracking_integration.py (3)

22-22: Verify the new expected value in the assertion.

The assertion has been changed to expect 8 tracks instead of 27. Ensure that this new expected value is correct based on the updated tracking logic or data.

---

44-44: Verify the tracker.run_tracker method invocation.

Ensure that the tracker object is correctly instantiated and that the run_tracker method handles the verbosity parameter appropriately.

---

51-51: Verify the tracker.run_tracker method invocation.

Ensure that the tracker object is correctly instantiated and that the run_tracker method handles the verbosity parameter appropriately.

tests/nn/test_tracker_components.py (2)

18-27: Verify the handling of the img_scale parameter and the test skipping logic.

Ensure that the img_scale parameter is correctly applied to the candidate_maker of the tracker and that the function appropriately skips tests when img_scale is not defined.

---

47-58: Verify the parameterized tests and the passing of the img_scale parameter.

Ensure that the parameterized tests are correctly implemented and that the img_scale parameter is passed appropriately to the run_tracker_by_name function.

sleap/config/pipeline_form.yaml (2)

458-463: New configuration option tracking.save_shifted_instances added.

This option is well-documented and the default value is set to true, which aligns with the PR summary indicating the enhancement of instance matching.

---

457-457: Verify the impact of removing tracking.max_tracking.

Ensure that the removal of this configuration option does not negatively impact any existing functionality or user workflows.

Verification successful

Verify the integration of tracking.max_tracks and check for any remaining references to tracking.max_tracking.

Ensure that tracking.max_tracks is correctly integrated into the codebase and that there are no remaining dependencies on tracking.max_tracking.

---

The integration of tracking.max_tracks is correct and there are no remaining dependencies on tracking.max_tracking.

The replacement of tracking.max_tracking with tracking.max_tracks has been successfully integrated into the configuration file, and there are no lingering references to the old configuration.

• tracking.max_tracks is correctly defined and used in the pipeline_form.yaml file.
• No remaining references to tracking.max_tracking were found in non-YAML files.

Scripts executed

The following scripts were executed for the analysis:

Script:

#!/bin/bash
# Description: Verify the impact of removing `tracking.max_tracking`.

# Test: Search for the usage of `tracking.max_tracking`. Expect: No occurrences.
rg --type yaml -A 5 $'tracking.max_tracking'

Length of output: 421

---

Script:

#!/bin/bash
# Description: Verify the integration of `tracking.max_tracks` and check for any remaining references to `tracking.max_tracking`.

# Test 1: Search for the usage of `tracking.max_tracks`. Expect: Relevant occurrences showing correct integration.
rg --type yaml -A 5 $'tracking.max_tracks'

# Test 2: Search for any remaining references to `tracking.max_tracking` in non-YAML files. Expect: No occurrences.
rg --type yaml -v $'tracking.max_tracking'

Length of output: 138158

sleap/gui/learning/dialog.py (1)

739-739: Enhanced robustness in run method.

The calculation of no_result_count now ensures it is non-negative, preventing potential errors in the inference process.

sleap/nn/tracking.py (7)

142-142: Verify the impact of changing the default value of save_shifted_instances.

Changing the default value of save_shifted_instances to True could impact memory usage and performance. Ensure that this change is intentional and tested.

---

532-548: LGTM!

The new method run_step encapsulates the logic for running a single step of tracking.

---

549-653: LGTM!

The new method run_tracker encapsulates the tracking logic and enhances reporting capabilities based on verbosity levels.

Tools

Ruff

614-614: Use enumerate() for index variable n_processed in for loop

(SIM113)

---

1707-1707: LGTM!

The modification aligns with the changes in the BaseTracker class.

---

520-521: LGTM! Verify the usage of new attributes.

The new attributes verbosity and report_rate enhance the reporting capabilities.

Ensure that these attributes are correctly used and tested throughout the codebase.

Verification successful

Let's modify the script to search for the usage of verbosity and report_rate in all Python files without specifying the file type.

---

The new attributes verbosity and report_rate are correctly used and tested throughout the codebase.

The attributes are integrated into the main functionality and are covered by tests, ensuring their proper usage.

• sleap/nn/tracking.py
• sleap/nn/inference.py
• tests/nn/test_tracking_integration.py
• tests/nn/test_inference.py
• tests/nn/test_tracker_components.py

Scripts executed

The following scripts were executed for the analysis:

Script:

#!/bin/bash
# Description: Verify the usage of new attributes `verbosity` and `report_rate`.

# Test: Search for the usage of `verbosity` and `report_rate`. Expect: Occurrences of the new attributes.
rg --type python 'verbosity|report_rate'

Length of output: 73

---

Script:

#!/bin/bash
# Description: Verify the usage of new attributes `verbosity` and `report_rate`.

# Test: Search for the usage of `verbosity` and `report_rate`. Expect: Occurrences of the new attributes.
rg 'verbosity|report_rate' --glob '*.py'

Length of output: 3311

---

742-747: LGTM! Verify the usage of new attributes.

The new attributes verbosity and report_rate enhance the reporting capabilities.

Ensure that these attributes are correctly used and tested throughout the codebase.

Verification successful

---

The new attributes verbosity and report_rate are correctly used and tested throughout the codebase.

The search results indicate that both attributes are utilized in various parts of the code, including the sleap/nn/tracking.py, sleap/nn/inference.py, and several test files. This confirms that the new attributes are integrated and tested.

• Files with relevant usage:
  • sleap/nn/tracking.py
  • sleap/nn/inference.py
  • tests/nn/test_tracking_integration.py
  • tests/nn/test_tracker_components.py
  • tests/nn/test_inference.py
  • sleap/gui/learning/runners.py
  • docs/notebooks/Training_and_inference_on_an_example_dataset.ipynb
  • docs/guides/cli.md

Scripts executed

The following scripts were executed for the analysis:

Script:

#!/bin/bash
# Description: Verify the usage of new attributes `verbosity` and `report_rate`.

# Test: Search for the usage of `verbosity` and `report_rate`. Expect: Occurrences of the new attributes.
rg --type python 'verbosity|report_rate'

Length of output: 73

---

Script:

#!/bin/bash
# Description: Verify the usage of new attributes `verbosity` and `report_rate`.

# Test: Search for the usage of `verbosity` and `report_rate` in Python files.
rg --type-add 'py:*.py' 'verbosity|report_rate'

Length of output: 5005

---

1484-1489: LGTM! Verify the usage of new attributes.

The new attributes verbosity and report_rate enhance the reporting capabilities.

Ensure that these attributes are correctly used and tested throughout the codebase.

Verification successful

---

The new attributes verbosity and report_rate are correctly used and tested throughout the codebase.

The attributes are integrated into multiple files, including sleap/nn/tracking.py, sleap/nn/inference.py, and various test files, ensuring their proper utilization and testing.

• Files with usage:
  • sleap/nn/tracking.py
  • sleap/nn/inference.py
  • tests/nn/test_tracker_components.py
  • tests/nn/test_inference.py
  • tests/nn/test_tracking_integration.py

Scripts executed

The following scripts were executed for the analysis:

Script:

#!/bin/bash
# Description: Verify the usage of new attributes `verbosity` and `report_rate`.

# Test: Search for the usage of `verbosity` and `report_rate`. Expect: Occurrences of the new attributes.
rg --type python 'verbosity|report_rate'

Length of output: 73

---

Script:

#!/bin/bash
# Description: Verify the usage of new attributes `verbosity` and `report_rate`.

# Test: Search for the usage of `verbosity` and `report_rate`. Expect: Occurrences of the new attributes.
rg --type py 'verbosity|report_rate'

Length of output: 3307

sleap/gui/widgets/video.py (2)

807-808: Initialization of _down_pos is appropriate.

The _down_pos attribute is correctly initialized to None to ensure it is defined before use.

---

1032-1032: Robustness improved with _down_pos check.

The check for _down_pos being None before comparing it with the current mouse position enhances the robustness of the click detection logic.

sleap/nn/inference.py (4)

48-51: Conditional import for cached_property is appropriate.

The conditional import ensures compatibility with Python versions below 3.8, which is a good practice.

---

163-165: Use of cached_property for report_period is appropriate.

The cached_property decorator improves performance by caching the computed value of report_period.

---

5341-5343: Inclusion of example_indices in LabelsReader initialization is beneficial.

The modification enhances the configurability by allowing more flexible handling of input frames based on user-defined indices.

---

5437-5440: Inclusion of progress_reporting in tracker instantiation is beneficial.

The modification allows for more detailed tracking feedback, enhancing the monitoring of the tracking process.

Files that changed from the base of the PR and between c37b715 and 138c225.

• sleap/nn/tracking.py (15 hunks)

sleap/nn/tracking.py (12)

4-5: Imports approved.

The additions of json, sys, and rich.progress are appropriate for the new reporting functionality.

---

520-521: New attributes verbosity and report_rate approved.

These attributes enhance user control over the output of the tracking process.

---

532-547: New run_step method approved.

This method clears tracks and processes instances for tracking.

---

742-746: New attributes verbosity and report_rate approved.

These attributes enhance user control over the output of the tracking process.

---

1484-1488: New attributes verbosity and report_rate approved.

These attributes enhance user control over the output of the tracking process.

---

1014-1015: New parameters progress_reporting and report_rate approved.

These parameters enhance user control over the output of the tracking process.

---

Line range hint 1226-1248:
Updated add_cli_parser_args method approved.

The method updates the CLI parser to include the new progress_reporting and report_rate parameters.

---

1280-1283: Updated get_by_name_factory_options method approved.

The method updates the factory options to include the new progress_reporting and report_rate parameters.

---

1484-1489: Updated track method approved.

The method enhances the tracking process by incorporating Kalman filters for better accuracy.

---

1707-1707: Modified retrack function approved.

The modification aligns with the removal of the standalone run_tracker function.

However, ensure that all function calls to retrack match the new implementation.

Verification successful

---

Modified retrack function approved.

The modification aligns with the removal of the standalone run_tracker function. The function calls to retrack match the new implementation.

• tests/nn/test_inference.py: Verified test case involving retrack.
• sleap/nn/tracking.py: Verified function definition and call.

Scripts executed

The following scripts were executed for the analysis:

Script:

#!/bin/bash
# Description: Verify all function calls to `retrack` match the new implementation.

# Test: Search for the function usage. Expect: Only occurrences of the new implementation.
rg --type python -A 5 $'retrack'

Length of output: 62

---

Script:

#!/bin/bash
# Description: Verify all function calls to `retrack` match the new implementation.

# Test: Search for the function usage. Expect: Only occurrences of the new implementation.
rg --type py -A 5 $'retrack'

Length of output: 1193

---

142-142: Verify the impact of changing the default value of save_shifted_instances to True.

This change could affect memory usage and performance.

Verification successful

Verify the impact of changing the default value of save_shifted_instances to True.

This change could affect memory usage and performance.

---

Verify the impact of changing the default value of save_shifted_instances to True.

The change in the default value of save_shifted_instances to True will increase memory usage due to the saving of shifted instances between frames. However, it is expected to improve instance matching. This behavior is documented and tested in the codebase.

• sleap/nn/tracking.py: Multiple references to save_shifted_instances show its impact on saving shifted instances.
• sleap/gui/learning/runners.py: Configuration references.
• tests/nn/test_inference.py: Test cases covering the new default behavior.

Scripts executed

The following scripts were executed for the analysis:

Script:

#!/bin/bash
# Description: Verify the impact of changing the default value of `save_shifted_instances` to `True`.

# Test: Search for the usage of `save_shifted_instances` in the codebase. Expect: Analyze its impact on memory and performance.
rg --type python -A 5 $'save_shifted_instances'

Length of output: 77

---

Script:

#!/bin/bash
# Description: Verify the impact of changing the default value of `save_shifted_instances` to `True`.

# Test: Search for the usage of `save_shifted_instances` in the codebase. Expect: Analyze its impact on memory and performance.
rg --type py -A 5 $'save_shifted_instances'

Length of output: 5528

---

549-653: New run_tracker method approved.

This method includes logic for different verbosity modes, providing visual feedback via a progress bar or JSON-formatted progress updates.

However, ensure that all function calls to run_tracker match the new signature.

Tools

Ruff

614-614: Use enumerate() for index variable n_processed in for loop

(SIM113)

Files that changed from the base of the PR and between 138c225 and 7d293c9.

• sleap/config/pipeline_form.yaml (2 hunks)
• sleap/gui/learning/dialog.py (1 hunks)
• sleap/gui/widgets/video.py (2 hunks)
• sleap/nn/inference.py (12 hunks)
• sleap/nn/tracking.py (24 hunks)
• sleap/util.py (2 hunks)
• tests/nn/test_tracker_components.py (2 hunks)
• tests/nn/test_tracking_integration.py (4 hunks)

• sleap/config/pipeline_form.yaml
• sleap/gui/learning/dialog.py
• sleap/gui/widgets/video.py
• sleap/nn/inference.py
• sleap/util.py

tests/nn/test_tracking_integration.py (3)

22-22: Verify the updated assertion.

The expected number of tracks has been reduced from 27 to 8. Ensure this change is correct and consistent with the updated tracking logic.

---

44-44: LGTM! Ensure correct initialization of tracker object.

The change to use tracker.run_tracker with verbosity improves encapsulation and maintainability.

---

51-51: LGTM! Ensure correct initialization of tracker object.

The change to use tracker.run_tracker with verbosity improves encapsulation and maintainability.

tests/nn/test_tracker_components.py (2)

18-28: LGTM! Improved flexibility and error handling.

The addition of the img_scale parameter and the use of pytest.skip() enhance the function's flexibility and robustness.

---

40-58: LGTM! Enhanced testing coverage.

The inclusion of img_scale as a parameter broadens the scope of the tests to include variations in image scaling.

sleap/nn/tracking.py (5)

520-531: LGTM! Improved user control and reporting.

The addition of verbosity and report_rate attributes, along with the enhanced run_tracker method, improve user control and reporting capabilities.

---

532-548: LGTM! New method run_step improves modularity.

The run_step method enhances modularity by encapsulating the logic for processing a single frame.

---

549-654: LGTM! Enhanced run_tracker method.

The run_tracker method now includes logic for different verbosity modes, providing better user feedback during the tracking process.

Tools

Ruff

614-614: Use enumerate() for index variable n_processed in for loop

(SIM113)

---

704-709: LGTM! Improved user control and reporting.

The addition of verbosity and report_rate attributes improves user control and reporting capabilities.

---

1009-1010: LGTM! Updated make_tracker_by_name method.

The make_tracker_by_name method now includes progress_reporting and report_rate parameters, enhancing user control over the tracking process.

Files that changed from the base of the PR and between 7d293c9 and 2be8b80.

• sleap/nn/tracking.py (23 hunks)
• tests/nn/test_tracker_components.py (2 hunks)
• tests/nn/test_tracking_integration.py (5 hunks)

tests/nn/test_tracking_integration.py (3)

22-22: Verify the change in the expected number of tracks.

The assertion change from 27 to 8 indicates a significant reduction in the number of tracks. Verify if this change aligns with the intended behavior and updated tracking logic.

---

45-45: LGTM! Verify the handling of the verbosity parameter.

The change to use tracker.run_tracker with an additional verbosity parameter is approved. Ensure that the verbosity parameter is handled correctly within the tracker class.

---

52-52: LGTM! Verify the handling of the verbosity parameter.

The change to use tracker.run_tracker with an additional verbosity parameter is approved. Ensure that the verbosity parameter is handled correctly within the tracker class.

tests/nn/test_tracker_components.py (2)

18-27: LGTM! Verify the handling of the new img_scale parameter.

The changes to rename the function and include the img_scale parameter enhance flexibility. Ensure that the new parameter and logic are correctly implemented.

---

40-40: LGTM! Verify the integration of the new img_scale parameter.

The changes to include the img_scale parameter in the tests enhance coverage. Ensure that the new parameter is correctly integrated into the tests.

sleap/nn/tracking.py (5)

141-141: Verify the change in the default value of save_shifted_instances.

The default value of save_shifted_instances is changed from False to True. Verify if this change aligns with the intended functionality and does not adversely impact memory usage and performance.

---

520-521: LGTM! Verify the implementation of new attributes and methods.

The new attributes (verbosity, report_rate) and methods (run_step, run_tracker) enhance the tracking process. Ensure that they are correctly implemented and integrated.

Also applies to: 527-531, 532-548, 549-654

---

704-709: LGTM! Verify the implementation of new attributes and modifications.

The new attributes (verbosity, report_rate) and modifications enhance the tracking process. Ensure that they are correctly implemented and integrated.

Also applies to: 742-747

---

1477-1482: LGTM! Verify the implementation of new attributes.

The new attributes (verbosity, report_rate) enhance the tracking process. Ensure that they are correctly implemented and integrated.

---

1699-1699: LGTM! Verify the modification to call tracker.run_tracker.

The change to call tracker.run_tracker aligns with the refactor. Ensure that the modification is correctly implemented.

Files that changed from the base of the PR and between 2be8b80 and 98c96dc.

• sleap/nn/inference.py (13 hunks)

• sleap/nn/inference.py

Files that changed from the base of the PR and between 98c96dc and 0bb7f9c.

• sleap/config/pipeline_form.yaml (2 hunks)
• sleap/gui/learning/dialog.py (1 hunks)
• sleap/gui/widgets/video.py (2 hunks)
• sleap/nn/inference.py (13 hunks)
• sleap/nn/tracking.py (23 hunks)
• sleap/util.py (2 hunks)
• tests/nn/test_tracker_components.py (2 hunks)
• tests/nn/test_tracking_integration.py (5 hunks)

tests/nn/test_tracking_integration.py (2)

45-45: Approve direct call to tracker.run_tracker.

The replacement of the run_tracker function call with a direct call to tracker.run_tracker(frames, verbosity="none") is consistent with the earlier changes and simplifies the code.

---

22-22: Verify the change in expected number of tracks.

The assertion for the number of tracks has been significantly reduced from 27 to 8. This change might be related to the improvements in tracking functionality mentioned in the PR objectives.

Could you provide more context on why the expected number of tracks has changed so drastically? This change might have implications for other parts of the system that depend on the number of tracks.

sleap/util.py (7)

6-6: LGTM: Good use of postponed annotation evaluation.

The addition of from __future__ import annotations is a good practice. It enables postponed evaluation of annotations, which can improve performance and reduce circular import issues.

---

16-16: LGTM: Improved import for type checking.

The addition of TYPE_CHECKING to the import statement is a good practice. It allows for conditional imports that are only evaluated during type checking, which can optimize runtime performance.

---

20-24: LGTM: Improved import handling and new feature support.

The try-except block for importing files ensures compatibility with different Python versions. The addition of rich.progress import supports the new RateColumn class functionality.

Also applies to: 30-30

---

36-38: LGTM: Fixed undefined name issue.

The conditional import of Task from rich.progress addresses the issue mentioned in the previous comment about the undefined name Task. It's now properly imported only during type checking.

---

40-48: LGTM: New feature for improved progress visualization.

The new RateColumn class is a great addition that aligns with the PR objective of improving progress visualization. It provides a custom rendering of the progress rate, displaying the speed in FPS.

---

Line range hint 315-315: LGTM: Improved path handling.

The modification to use files("sleap").joinpath(filename) is a more modern and flexible approach for creating the package file path.

---

Line range hint 320-321: LGTM: Enhanced config file handling.

The additions to the get_config_file function provide more flexibility in config file handling. The new parameters ignore_file_not_found and get_defaults, along with the associated logic, allow for ignoring file not found errors and retrieving default configs. This aligns well with the PR objective of improving default options handling.

Also applies to: 332-333, 340-341, 349-350

sleap/config/pipeline_form.yaml (3)

458-463: New option to save shifted instances during tracking

The addition of the tracking.save_shifted_instances parameter aligns with the PR objectives of improving tracking functionality. This new option allows users to save flow-shifted instances between frames, which can enhance instance matching at the cost of increased memory usage.

This change improves the flexibility of the tracking process and gives users more control over the trade-off between accuracy and memory usage.

---

458-463: Summary of changes to tracking configuration

The modifications to the tracking configuration in this file align with the PR objectives of improving tracking functionality. The addition of the save_shifted_instances option provides users with more control over the tracking process, allowing for a trade-off between improved instance matching and memory usage.

However, the removal of the max_tracks parameter might affect existing users who relied on this option. It's important to ensure that this change doesn't break any existing functionality and that appropriate alternatives are provided if necessary.

These changes enhance the flexibility of the tracking process. However, please consider the following actions:

1. Update the user documentation to reflect these changes, especially regarding the new save_shifted_instances option and its performance implications.
2. Verify that the removal of max_tracks doesn't negatively impact existing users or workflows.
3. If max_tracks functionality is still needed, consider implementing it in a different way or providing clear migration instructions for users.

---

458-463: Verify the impact of saving shifted instances on memory usage

While the new save_shifted_instances option can improve instance matching, it's important to ensure that the increased memory usage doesn't cause issues in resource-constrained environments.

Let's check if there are any memory-related comments or considerations in the codebase:

sleap/gui/learning/dialog.py (2)

736-736: Approve: Prevent negative count with max function

The modification to use max(0, total_count - new_counts) is a good improvement. It ensures that no_result_count is never negative, which could happen if new_counts somehow exceeded total_count. This change adds robustness to the code and prevents potential issues or confusing output related to negative frame counts.

---

736-736: Verify similar calculations in the codebase

While this change is good, it might be worth checking if there are similar calculations elsewhere in the codebase that could benefit from the same robustness check.

Let's search for similar patterns in the codebase:

sleap/nn/tracking.py (7)

4-9: Imports updated to support new functionality.

The new imports (json, sys, time) are added to support the enhanced tracking and reporting features. The removal of the unused defaultdict import aligns with the previous review comment and improves code cleanliness.

---

33-34: New import for progress reporting.

The addition of RateColumn from sleap.util supports the enhanced progress reporting feature in the tracking system.

---

141-141: Default value for save_shifted_instances changed to True.

This change aligns with the PR objectives, which mentioned changing the default value for the save_shifted_instances option to True. This modification will likely improve the default behavior of instance tracking.

---

527-531: New cached_property for report_period added.

The addition of the report_period cached_property provides a clean and efficient way to calculate the time interval between progress reports. This improves code readability and potentially performance by caching the calculated value.

---

1085-1086: Progress reporting parameters added to make_tracker_by_name method.

The addition of progress_reporting and report_rate parameters to the make_tracker_by_name method allows for customization of the progress reporting features when creating a tracker instance. This change provides flexibility in how tracking progress is reported, aligning with the enhanced progress reporting capabilities introduced in this update.

---

Line range hint 1-1699: Summary of changes and overall assessment.

The modifications in this file significantly enhance the progress reporting capabilities of the tracking system, aligning well with the PR objectives. Key improvements include:

1. Addition of verbosity modes for flexible progress reporting.
2. Implementation of a rich progress bar and JSON output options.
3. Consistent updates across different tracker implementations (BaseTracker, Tracker, KalmanTracker).
4. New methods for step-by-step tracking with progress updates.

These changes will greatly improve the user experience by providing better feedback during the tracking process. The code is generally well-structured and consistent, with only minor suggestions for improvements in documentation and input validation.

The overall changes are well-implemented and will positively impact the SLEAP application's tracking functionality.

Tools

Ruff

614-614: Use enumerate() for index variable n_processed in for loop

(SIM113)

---

614-614: False positive from static analysis tool.

The static analysis tool (Ruff) suggests using enumerate() for the index variable n_processed in the for loop. However, this suggestion is not applicable in this context. The variable n_processed is not used as an index for the loop, but rather as a counter of processed frames. The current implementation is correct and more efficient for its purpose.

The current implementation is appropriate for the intended functionality.

Tools

Ruff

614-614: Use enumerate() for index variable n_processed in for loop

(SIM113)

sleap/gui/widgets/video.py (3)

819-820: LGTM: Good addition for mouse event handling.

The new instance variable _down_pos is a good practice for tracking the initial position of the mouse when a button is pressed. This will allow for more precise control over mouse events, such as distinguishing between clicks and drags.

---

1044-1045: LGTM: Improved mouse movement detection.

The updated logic for detecting mouse movement during a click is more robust. By checking if _down_pos is not None before comparing it with the current position, we prevent potential errors that could occur if _down_pos hasn't been set (e.g., if mouseReleaseEvent is somehow triggered without a corresponding mousePressEvent).

---

Line range hint 1-1185: Overall: Good improvements to mouse event handling.

The changes in this file are focused on improving the mouse event handling, particularly in the mouseReleaseEvent method. The addition of the _down_pos instance variable and its use in detecting mouse movement during clicks enhance the robustness of the user interaction logic.

These modifications should lead to more reliable distinction between clicks and drags, improving the overall user experience without introducing unnecessary complexity.

sleap/nn/inference.py (6)

49-53: LGTM: Conditional import of cached_property.

This change adds compatibility for Python versions below 3.8 by conditionally importing cached_property. This is a good practice for maintaining backwards compatibility.

---

415-417: Improved ETA calculation.

Compiling the loop examples before starting the timer is a good optimization for improving the accuracy of the ETA calculation. This change addresses the PR objective of enhancing the ETA calculation in the GUI progress bar.

---

5388-5390: Updated LabelsReader initialization.

The change in the LabelsReader initialization adds the example_indices parameter, which is derived from frame_list(args.frames). This modification allows for more flexible frame selection during inference, addressing the PR objective of adding functionality to run tracking only on defined frames.

---

5484-5487: Added progress_reporting to tracker initialization.

The addition of the progress_reporting parameter to the tracker initialization is a good improvement. It allows for consistent progress reporting between the predictor and the tracker.

---

5655-5660: Improved frame filtering for tracking.

This change improves the efficiency of frame filtering when running tracking on an existing prediction file. By converting the example_indices to a set, the search operation becomes O(1) instead of O(n), which can significantly speed up the process for large datasets.

---

5684-5684: Use of os.fspath for consistent path handling.

The use of os.fspath(data_path) is a good practice for ensuring consistent path handling across different operating systems. This change improves the robustness of the code.

Files that changed from the base of the PR and between 0bb7f9c and ed8d14b.

• sleap/nn/inference.py (11 hunks)
• sleap/nn/tracking.py (23 hunks)

sleap/nn/tracking.py (9)

141-141: Approved: Default value for save_shifted_instances changed to True.

The default value has been appropriately changed from False to True. As mentioned in the PR objectives, this restores access to the save_shifted_instances setting in the GUI, which was removed in a previous commit.

Saving the shifted instances between elapsed frames uses a bit more memory but leads to better instance matches.

---

520-521: Approved: New verbosity and report_rate attributes added to BaseTracker.

The addition of the verbosity and report_rate attributes to the BaseTracker class aligns with the PR objective of introducing different verbosity modes for progress reporting during tracking.

• verbosity likely controls the mode of progress reporting (e.g., "rich", "json", "none").
• report_rate likely controls the frequency of progress updates.

These attributes will enable more flexible and informative progress reporting during the tracking process.

---

527-530: Approved: New report_period property added to BaseTracker.

The addition of the report_period property to the BaseTracker class is a useful enhancement. It provides a convenient way to access the time interval between progress reports, calculated based on the report_rate attribute.

Using the @cached_property decorator ensures that the value is calculated only once and cached for subsequent accesses, which can improve performance by avoiding redundant calculations.

---

549-597: Approved: New _run_tracker_json method added to BaseTracker.

The addition of the _run_tracker_json method to the BaseTracker class is a significant enhancement that implements the "json" verbosity mode for progress reporting during tracking. This aligns with the PR objective of introducing different verbosity modes for progress reporting.

Key features of the _run_tracker_json method:

• Iterates over the frames, calls run_step on each frame, and yields the new labeled frame.
• Tracks the progress and timing information, including the number of processed frames, total frames, elapsed time, processing rate, and estimated time remaining (ETA).
• Reports the progress in JSON format, which can be easily captured and parsed for programmatic monitoring.
• Uses the max_length argument to limit the length of the deques used for calculating the recent processing rate.

This method will greatly enhance the ability to monitor the progress of the tracking process programmatically, especially in scenarios like headless servers or notebooks.

Tools

Ruff

568-568: Use enumerate() for index variable n_processed in for loop

(SIM113)

---

599-623: Approved: New _run_tracker_rich method added to BaseTracker.

The addition of the _run_tracker_rich method to the BaseTracker class is another significant enhancement that implements the "rich" verbosity mode for progress reporting during tracking. This aligns with the PR objective of introducing different verbosity modes for progress reporting.

Key features of the _run_tracker_rich method:

• Uses the rich library to display a visually appealing and informative progress bar during tracking.
• Iterates over the frames, calls run_step on each frame, and yields the new labeled frame.
• Updates the progress bar with the current progress percentage, estimated time remaining (ETA), and processing rate.
• Manually refreshes the progress bar based on the report_period to support notebooks.

This method will greatly improve the user experience by providing a clear and informative visual representation of the tracking progress, especially when running in interactive environments like notebooks.

---

625-661: Approved: Updated run_tracker method in BaseTracker.

The updates to the run_tracker method in the BaseTracker class are a significant improvement that incorporates the new verbosity modes for progress reporting during tracking. This aligns with the PR objective of introducing different verbosity modes for progress reporting.

Key changes in the updated run_tracker method:

• Accepts frames, verbosity, and final_pass as arguments.
• Checks the is_valid property to determine if tracking should be performed.
• Uses the verbosity argument or the verbosity attribute to determine the progress reporting mode.
• Delegates the actual tracking and progress reporting to the appropriate _run_tracker_* method based on the verbosity mode.
• Supports running a final pass after tracking if final_pass is True.

These updates provide flexibility and modularity in the tracking process, allowing users to choose the desired verbosity mode and enabling the integration of different progress reporting mechanisms.

---

750-754: Approved: Updated verbosity and added report_rate attributes in Tracker.

The changes to the verbosity and report_rate attributes in the Tracker class are in line with the PR objective of introducing different verbosity modes for progress reporting during tracking.

• The verbosity attribute now has a default value of "none", meaning that by default, no progress reporting will be displayed during tracking.
• The validator for the verbosity attribute ensures that only valid verbosity modes ("none", "rich", "json") can be assigned.
• The addition of the report_rate attribute with a default value of 2.0 allows controlling the frequency of progress updates during tracking.

These changes provide flexibility in configuring the progress reporting behavior of the Tracker class, enabling users to choose the desired verbosity mode and update frequency.

---

1228-1233: Verify the data type consistency for the save_shifted_instances option.

The default value for the save_shifted_instances option in the Tracker.get_by_name_factory_options method has been changed from False to 1. This change aligns with the modification made to the save_shifted_instances attribute in the FlowCandidateMaker class, as mentioned in the PR objectives.

However, there seems to be an inconsistency in the data type used for the default value. The option's data type is specified as int, but using 1 instead of True might cause confusion.

Please ensure that the data type for the save_shifted_instances option is consistently used throughout the codebase. If the intended data type is bool, consider updating the default value to True for clarity.

---

1247-1250: Approved: Updated help text for the kf_init_frame_count option.

The changes to the help text for the kf_init_frame_count option in the Tracker.get_by_name_factory_options method provide clearer guidance on the behavior of this option.

By explicitly mentioning that setting kf_init_frame_count to 0 means no Kalman filters will be used, the updated help text improves the understanding of this option's functionality and makes it easier for users to configure the tracking behavior correctly.

sleap/nn/inference.py (4)

49-53: LGTM!

The code changes are approved. Conditionally defining cached_property based on the Python version is a good way to maintain compatibility.

---

164-167: The existing discussion in the comments is still relevant. No need to duplicate the reasoning for using @cached_property here.

---

371-407: LGTM!

The new _process_batch method looks good. It nicely encapsulates the batch processing logic and correctly handles updating the input dictionary with model predictions and adjusting for potential scaling.

---

Line range hint 408-523: Looks good!

The new _run_batch_json and _run_batch_rich methods are a nice addition to provide different progress reporting modes. The implementation looks solid, correctly tracking progress, timing and reporting.

Updating _predict_generator to use these methods based on the verbosity setting is the right approach.

Files that changed from the base of the PR and between ed8d14b and 854432a.

• sleap/nn/inference.py (11 hunks)
• sleap/nn/tracking.py (23 hunks)

sleap/nn/tracking.py (10)

532-661: New run_tracker method added with enhanced progress reporting.

The run_tracker method is a significant addition that implements comprehensive progress reporting for the tracking process. It supports different verbosity modes ("rich", "json", and default) as mentioned in the PR objectives. This enhancement will greatly improve user experience by providing better feedback during the tracking process.

Key improvements:

1. Support for rich progress bar display
2. JSON output option for programmatic monitoring
3. Flexible reporting rate

Consider the following improvements:

1. Extract the progress reporting logic for each verbosity mode into separate methods to improve readability and maintainability.
2. Add error handling for potential exceptions during the tracking process.
3. Consider using a context manager for the "json" verbosity mode to ensure proper resource management, similar to the "rich" mode.

Tools

Ruff

568-568: Use enumerate() for index variable n_processed in for loop

(SIM113)

---

599-623: LGTM!

The _run_tracker_rich method effectively utilizes the rich library to display an informative and visually appealing progress bar for the "rich" verbosity mode. The code is well-structured and follows best practices.

---

750-754: LGTM!

The addition of the verbosity and report_rate attributes to the Tracker class aligns with the enhanced progress reporting feature. The validation of the verbosity attribute ensures that only valid verbosity modes are accepted, and the default value of report_rate is reasonable.

---

1305-1319: LGTM!

The changes to the add_cli_parser_args method are necessary to support the new "tracker" option without a default value, allowing for the detection of malformed command-line arguments. The addition of the other command-line arguments with their default values is consistent with the existing code.

---

Line range hint 1017-1094: LGTM!

The changes to the make_tracker_by_name method are consistent with the enhanced progress reporting feature. The new progress_reporting argument allows for specifying the desired verbosity mode when creating a Tracker instance, and the verbosity and report_rate attributes are set accordingly. The changes are well-integrated with the existing code.

---

1228-1233: LGTM!

The change in the default value of the save_shifted_instances option from False to True aligns with the PR objective of restoring access to this setting in the GUI. The updated help text accurately reflects the change in default value and provides clarity on the behavior of the option. The change is consistent with the overall goals of the PR.

---

1247-1250: LGTM!

The updated help text for the kf_init_frame_count option provides clearer guidance on the behavior of the option. It explicitly states that a value of 0 indicates that no Kalman filters will be used. This clarification improves the usability and understanding of the option. The change is minor but beneficial.

---

1271-1274: LGTM!

The updated help text for the oks_score_weighting option provides a clearer explanation of the behavior when the value is set to 0 or 1. It specifies that a value of 0 (default) means only the distance between the reference and query keypoint is used to compute the similarity, while a value of 1 means each distance is weighted by the prediction scores of the reference and query keypoint. This clarification enhances the understanding and usability of the option. The change is minor but valuable.

---

1282-1285: LGTM!

The updated help text for the oks_normalization option provides a detailed explanation of the behavior for different values. It clarifies that a value of "all" means the similarity score is normalized by the number of reference points, "ref" means the score is normalized by the number of visible reference points, and "union" means the score is normalized by the number of points both visible in query and reference instance. This detailed explanation improves the understanding and usability of the option. The change is valuable for users to make informed decisions when configuring the tracker.

---

1485-1489: LGTM!

The addition of the verbosity and `report_

sleap/nn/inference.py (4)

49-53: LGTM!

The code adds compatibility for Python versions below 3.8 by conditionally defining cached_property. This is a good practice to ensure backward compatibility.

---

164-167: Good optimization based on learnings!

The change from @property to @cached_property for report_period is a good optimization. As mentioned in the learnings:

When deciding between @cached_property and @lru_cache, prefer @cached_property when the goal is to ensure that the cached value is automatically deleted when the instance is deleted, avoiding potential memory leaks.

This ensures that the cached value is automatically deleted when the instance is deleted, avoiding potential memory leaks.

---

371-407: Improved modularity and clarity!

The new _process_batch method encapsulates the logic for running inference on a batch and updating the input dictionary with predictions. This improves modularity and clarity by extracting the inline processing logic into a dedicated method. The method handles important postprocessing steps such as:

• Skipping inference if the model is not loaded
• Running inference on the current batch
• Adding model outputs to the input data example
• Converting video_ind and frame_ind to numpy arrays if they are TensorFlow tensors
• Adjusting instance_peaks based on potential SizeMatcher scaling

The code is well-structured and follows good practices.

---

508-523: Improved ETA calculation and flexible progress reporting!

The changes introduce several improvements to the inference loop:

1. Compiling loop examples before starting the timer improves the ETA calculation by providing a more accurate estimate of the total number of examples. This enhances the user experience by providing a more reliable progress estimate.

2. The optional progress reporting based on the verbosity attribute allows for flexibility in how progress is displayed:

   • The "rich" option displays an informative and visually appealing progress bar using the _run_batch_rich method.
   • The "json" option outputs progress information as JSON using the _run_batch_json method, which can be useful for programmatic monitoring or logging.
   • The "none" option allows for silent execution without any progress reporting, suitable for scenarios where progress updates are not needed.

The code is well-organized and uses appropriate methods for different verbosity levels, making it easy to understand and maintain.

Files that changed from the base of the PR and between 854432a and 25cd34b.

• sleap/config/pipeline_form.yaml (2 hunks)
• sleap/gui/learning/dialog.py (1 hunks)
• sleap/gui/widgets/video.py (2 hunks)
• sleap/nn/inference.py (11 hunks)
• sleap/nn/tracking.py (23 hunks)
• sleap/util.py (2 hunks)
• tests/nn/test_tracker_components.py (2 hunks)
• tests/nn/test_tracking_integration.py (5 hunks)

• sleap/config/pipeline_form.yaml
• sleap/gui/learning/dialog.py
• sleap/gui/widgets/video.py
• sleap/util.py

tests/nn/test_tracking_integration.py (4)

22-22: Verify the change in the expected number of tracks.

The assertion in the test_simple_tracker function has been updated to expect 8 tracks instead of 27. This change suggests a potential adjustment in the underlying tracking logic or the data being processed.

Please confirm that this change aligns with the intended behavior and provide a brief explanation for the adjustment in the expected number of tracks.

---

45-45: Verify the impact of removing the run_tracker function.

The run_tracker function has been removed, and its functionality has been replaced by a direct call to tracker.run_tracker(frames, verbosity="none") in the make_ground_truth function. While this change simplifies the code, it's important to ensure that it doesn't introduce any unintended side effects or break existing functionality.

Please confirm that the removal of the run_tracker function and the direct utilization of the tracker's method have been thoroughly tested and do not negatively impact the overall behavior of the code.

---

52-52: Verify the impact of removing the run_tracker function in the compare_ground_truth function.

Similar to the change in the make_ground_truth function, the run_tracker function has been removed, and its functionality has been replaced by a direct call to tracker.run_tracker(frames, verbosity="none") in the compare_ground_truth function. While this change simplifies the code, it's important to ensure that it doesn't introduce any unintended side effects or break existing functionality.

Please confirm that the removal of the run_tracker function and the direct utilization of the tracker's method in the compare_ground_truth function have been thoroughly tested and do not negatively impact the overall behavior of the code.

---

131-133: LGTM!

The modification in the make_filename function, where the string concatenation for the file path has been replaced with os.path.join, improves the robustness of file path handling across different operating systems. This change enhances the portability and maintainability of the code.

tests/nn/test_tracker_components.py (3)

18-32: LGTM!

The new run_tracker_by_name function is well-structured and handles the img_scale parameter appropriately. Skipping the tracking process when img_scale is not supported is a good practice to prevent unnecessary execution. The assertion at the end ensures that the tracking process produces the expected number of frames.

---

40-40: Also applies to: 47-47, 53-58

---

Line range hint 77-84:

sleap/nn/tracking.py (11)

625-661: New run_tracker method added with enhanced progress reporting.

The run_tracker method is a significant addition that implements comprehensive progress reporting for the tracking process. It supports different verbosity modes ("rich", "json", and default) as mentioned in the PR objectives. This enhancement will greatly improve user experience by providing better feedback during the tracking process.

---

599-623: LGTM!

The _run_tracker_rich method is well-implemented and provides a rich progress bar for tracking progress. It uses a context manager to ensure proper resource management.

---

532-547: LGTM!

The run_step method is correctly implemented and clears the tracks before running the tracking step.

---

527-530: LGTM!

The report_period cached property is correctly implemented and uses the report_rate attribute to calculate the report period.

---

750-754: LGTM!

The new verbosity and report_rate attributes are correctly added to the Tracker class and will be used for progress reporting during tracking.

---

1017-1018: LGTM!

The make_tracker_by_name class method correctly handles the new progress_reporting and report_rate arguments and passes them to the Tracker constructor.

---

1228-1233: LGTM!

The get_by_name_factory_options class method correctly updates the default value of save_shifted_instances to 1. This change aligns with the PR objective of restoring access to this setting in the GUI.

---

1305-1312: LGTM!

The add_cli_parser_args class method correctly handles the case when the default value is defined for "tracking.tracker" by not passing the default value to the argument parser. This change prevents the detection of malformed command line arguments.

---

1485-1489: LGTM!

The new verbosity and report_rate attributes are correctly added to the KalmanTracker class and will be used for progress reporting during tracking.

---

1707-1707: LGTM!

The run_tracker method of the KalmanTracker class correctly passes the frames argument to the init_tracker.run_tracker method.

---

568-568: Skip the Ruff suggestion.

The suggestion to use enumerate() for index variable n_processed is not applicable in this case because n_processed is not used as an index variable in the for loop. It is a counter variable that is incremented in each iteration of the loop. Using enumerate() would not provide any benefit here.

Tools

Ruff

568-568: Use enumerate() for index variable n_processed in for loop

(SIM113)

sleap/nn/inference.py (4)

164-167: LGTM!

The code changes are approved. Using @cached_property is the right choice here to avoid potential memory leaks.

---

371-407: Looks good!

The new _process_batch function correctly handles running inference on a batch and postprocessing the results. The logic for skipping inference when the model is not loaded, running inference, updating the input dictionary, and adjusting for SizeMatcher scaling is implemented properly.

---

408-455: LGTM!

The new _run_batch_json function correctly processes batches of examples and reports the progress in JSON format. The logic for tracking timing and progress, computing the recent rate and ETA, and printing the JSON progress information at the specified report period is implemented properly.

---

456-486: Looks good!

The new _run_batch_rich function correctly processes batches of examples and reports the progress using the rich progress bar. The setup of the rich progress bar with custom configuration, updating the progress for each batch, and manually refreshing the progress bar to support notebooks is implemented properly.

Files that changed from the base of the PR and between 10799b9 and 809e580.

• sleap/config/pipeline_form.yaml (1 hunks)
• sleap/nn/inference.py (10 hunks)
• sleap/nn/tracking.py (13 hunks)
• sleap/util.py (1 hunks)
• tests/nn/test_tracker_components.py (3 hunks)
• tests/nn/test_tracking_integration.py (4 hunks)

tests/nn/test_tracking_integration.py (4)

5-5: LGTM: Import Path from pathlib

The addition of from pathlib import Path is a good change. It aligns with the project's move towards using pathlib for cross-platform path handling.

---

46-46: LGTM: Simplified tracker execution

The direct use of tracker.run_tracker(frames, verbosity="none") in both make_ground_truth and compare_ground_truth functions is a good simplification. It removes the unnecessary intermediate run_tracker function, making the code more straightforward.

Also applies to: 53-53

---

134-136: LGTM: Improved path handling

The use of Path.joinpath for constructing the file path is an excellent change. It improves cross-platform compatibility and follows the project's move towards using pathlib for path handling.

---

23-23: Verify the change in expected number of tracks

The assertion for the number of tracks has been significantly reduced from 27 to 8. This change might reflect improvements in the tracking algorithm or changes in its configuration.

Please confirm:

1. Is this change intentional?
2. What improvements or changes in the tracking algorithm led to this reduction?
3. Are there any implications for the overall tracking performance?

sleap/util.py (2)

29-29: LGTM: New imports for progress tracking and type hinting.

The addition of rich.progress import and the conditional import of Task for type hinting are appropriate for the new RateColumn class and improve type checking without affecting runtime performance.

Also applies to: 36-37

---

1-49: Summary: Enhancements align well with PR objectives.

The changes in this file, particularly the addition of the RateColumn class, contribute significantly to the PR's goal of improving tracking functionality and progress display in the SLEAP application. These modifications lay the groundwork for implementing the progress bar during tracking, as outlined in the PR objectives.

The new imports and class definition are well-structured and follow best practices. With the suggested minor improvements (removing unused imports and adding a clarifying comment), this file will be in excellent shape to support the enhanced tracking features.

🧰 Tools

🪛 Ruff

8-8: base64 imported but unused

Remove unused import: base64

(F401)

---

31-31: PIL.Image imported but unused

Remove unused import: PIL.Image

(F401)

tests/nn/test_tracker_components.py (2)

18-32: Function run_tracker_by_name implemented correctly

The new run_tracker_by_name function correctly replaces tracker_by_name and appropriately handles the img_scale parameter.

---

50-61: Parameters added to test_tracker_by_name appropriately

The addition of img_scale and count parameters to test_tracker_by_name is appropriate and correctly integrated into the test function.

sleap/nn/tracking.py (5)

4-8: LGTM on added imports for progress reporting functionality

The newly added imports are appropriate and necessary for the implementation of progress reporting features.

---

12-13: Importing rich.progress enhances progress bar capabilities

The addition of import rich.progress is essential for the rich progress bar implementation in the tracking process.

---

602-627: New method _run_tracker_rich adds rich progress bar display

The _run_tracker_rich method introduces a rich progress bar during tracking, significantly improving user experience by providing visual feedback.

---

552-601: New method _run_tracker_json provides JSON-based progress reporting

The implementation of the _run_tracker_json method effectively adds JSON-based progress reporting, enabling programmatic monitoring of the tracking process.

🧰 Tools

🪛 Ruff

571-571: Use enumerate() for index variable n_processed in for loop

(SIM113)

---

628-665: Updates to run_tracker method to incorporate verbosity options

The run_tracker method now supports different verbosity levels, allowing for flexible progress reporting tailored to various use cases.

sleap/nn/inference.py (1)

508-509: Confirm that data_provider supports len()

In the line n_total = len(data_provider), ensure that data_provider implements the __len__ method. If not, this will raise a TypeError. Verify that data_provider supports the length operation.

Files that changed from the base of the PR and between 809e580 and 2e9b45a.

• sleap/nn/tracking.py (14 hunks)
• tests/nn/test_tracker_components.py (3 hunks)
• tests/nn/test_tracking_integration.py (4 hunks)

• tests/nn/test_tracker_components.py
• tests/nn/test_tracking_integration.py

sleap/nn/tracking.py (14)

4-9: Imports are appropriately added for new functionality.

The imports of json, sys, deque, time, and typing annotations are necessary for the added methods and enhancements.

---

12-13: Necessary imports numpy and rich.progress added.

These imports are required for numerical computations and enhanced progress reporting.

---

35-36: Import of RateColumn for progress bar display.

The RateColumn is used in the rich progress bar to display processing rates.

---

523-525: Duplicate comment: Missing default values and validators for verbosity and report_rate.

---

530-534: Duplicate comment: Potential division by zero in report_period due to report_rate being zero.

---

553-602: The _run_tracker_json method is correctly implemented for JSON progress reporting.

The method effectively handles progress calculations and outputs, providing useful feedback during tracking.

🧰 Tools

🪛 Ruff

572-572: Use enumerate() for index variable n_processed in for loop

(SIM113)

---

603-628: The _run_tracker_rich method utilizes rich library effectively for progress display.

The implementation is appropriate and follows best practices for progress bar integration.

---

630-666: run_tracker method incorporates verbosity options and handles execution flow correctly.

The method correctly delegates to the appropriate progress reporting method based on the verbosity setting.

---

755-760: Added verbosity and report_rate with validation in Tracker class.

Including validators and default values ensures robust usage of these attributes.

---

1029-1030: Duplicate comment: Inconsistent default verbosity values across classes.

---

1105-1106: Passing verbosity and report_rate to the tracker instance.

This ensures that the configured verbosity and report rate are correctly applied.

---

1497-1502: Added verbosity and report_rate with validation in KalmanTracker class.

Consistent implementation with the Tracker class.

---

1566-1566: The track method in KalmanTracker correctly implements the abstract base class signature.

Ensures compatibility with the BaseTracker.

---

1720-1720: Updated retrack function to use run_tracker method.

This change aligns the main function with the new tracking workflow.