Code_EvaDB

Python packages:

1. Install EvaDB and norfair in python environment

2. Install some other packages for color_NN model of AICITY according to requirements.txt

Checkpoints:

1. Download color_NN model, and save it like 1_query_stateless_property/data/checkpoints/CLIP_recognition_color/checkpoint_recognition_color.pth

Video data:

1. Download Banff, Jackson and Southampton, and put them in directory Three_Datasets. Ensure that Three_Datasets and Code_EvaDB are at the same level.

2. Download ua_detrac.mp4 as wget -nc "https://www.dropbox.com/s/k00wge9exwkfxz6/ua_detrac.mp4?raw=1" -O ua_detrac.mp4, and put it in 1_query_stateless_property/

How to run:

1. Example for query red cars:

   • cd ./1_query_stateless_property/
   • mkdir result
   • python query_red_car.py

2. Move the bash file eval.sh to the parent directory of root directory Code_EvaDB, which means that eval.sh and the root directory Code_EvaDB are at the same level. Then use: bash eval.sh.