First, cd
into /home/aistore17/Final_submission/utils/final
. Then run
python3 main.py --weights /home/aistore17/results/yolov7_400ep.pt -cfc 0.93 -cfs 0.8 -nfa 10 -msc 3 \
--out-path ../../results --source /home/aistore17/Datasets/4.TestVideosSample-2/1
Running this will save a video with change log written on top of center camera video and a csv file of change log with its corresponding time. Default parameters output the best results.
Change --source argument to /home/aistore17/Datasets/4.TestVideosSample-2/{2~6}
to run video sample 2~6.
Source folder has to have ONLY 5 camera videos in it with the format of #_center.mp4
, #_left_back.mp4
,
#_left_front.mp4
, #_right_back.mp4
, #_right_front.mp4
. # indicates the number of video.
The output file path is /home/aistore17/Final_submission/results
There are some more arguments you can use to run main.py
- --weights : type=str, default = '/home/aistore17/results/yolov7_400ep.pt', yolo model to use
- --iou-thres : type=float, default=0.45
- --source : type=str, default='/home/aistore17/Datasets/4.TestVideosSample-2/1', help='0 for webcams, else directory path to videos'
- --img-size : type=int, default=640, help='inference size (pixels)'
- --output-path : type=str, default='../../results', help='path to save results'
- --save-vid : type=bool, default=True, help='whether to save video as output'
- --save-csv : type=bool, default=True, help='whether to save csv file as output'
- -cfc, --conf-thres-center : type=float, default=0.25, confidence threshold of center camera
- -cfs, --conf-thres-side : type=float, default=0.25, confidence threshold of side cameras
- -nfa, --num-frames-to-avg : type=int, default=5, help='number of frames to average'
- -msc, --max-stable-count : type=int, default=5, help='maximum counts for results to stay stable'