Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Conversation

@headupinclouds
Copy link
Contributor

  • fix CV_8UC1 thresholds for direct cv::Mat4b gpgpu ACF texture processing (these were being scaled (x 255) correctly during *.mat and *.cpb deserialization but they also required a conversion to CV_8UC1 format — CV_Assert() tests have been added to cover this case
  • various updates to the pipeline test to support direct in report testsing and comparison between the ACF cpu pyramid based processing and the optimized GPGPU pyramid approximation
    • refactoring of GPUDetectionPipeline to accomodate both CPU and interleaved read/rocessing GPU based pipelines
    • return backing cv::Mat storage for FrameInput in pipeline Application test : getFrameInput(ogles_gpgpu::FrmaeInput &frame)
    • add pipeline GPUDetectionPipeline flags to toggle cpu/gpu processing for comparison/debugging
    • move VideoCaptureImage.{h,cpp} to separate files
    • add --size=x command line option since there is no generic approach for selecting either max or other valid camera resolution using cv::VideoCapture
  • add simple parallel_for evaluation of detection scales (use randomized indices for first scheduler)
  • bump version to v0.1.8
  • clang-format everything

* fix CV_8UC1 thresholds for direct cv::Mat4b gpgpu ACF texture processing (these were being scaled (x 255) correctly during `*.mat` and `*.cpb` deserialization but they also required a conversion to CV_8UC1 format — CV_Assert() tests have been added to cover this case
* various updates to the pipeline test to support direct in report testsing and comparison between the ACF cpu pyramid based processing and the optimized GPGPU pyramid approximation
  + refactoring of GPUDetectionPipeline to accomodate both CPU and interleaved read/rocessing GPU based pipelines
  + return backing cv::Mat storage for FrameInput in pipeline Application test : getFrameInput(ogles_gpgpu::FrmaeInput &frame)
  + add pipeline GPUDetectionPipeline flags to toggle cpu/gpu processing for comparison/debugging
  + move VideoCaptureImage.{h,cpp} to separate files
* add simple parallel_for evaluation of detection scales (use randomized indices for first scheduler)
* bump version
…ture (--size=1920x1080), since auto/max approach doesn't work consistently
@headupinclouds headupinclouds merged commit d821ad3 into master May 16, 2018
@headupinclouds headupinclouds deleted the pr.gpgpu.threshold branch May 16, 2018 17:00
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants