Closed abhaygargab closed 3 years ago
@abhaygargab Hi, can you please specify the format of this annotation just for this frame? I suppose it is class_id track_id x y w h Do let me know because I am currently working in multi-class multi object tracking for a procedural task(50 salads dataset). I am unable to figure out @CaptainEven 's code in order to apply it on my target dataset
Hi @Ravsehajsinghpuri , Please follow the installation steps and data preparation from https://github.com/ifzhang/FairMOT. And follow this issue #12 to understand how to modify the dataset for MCMOT.
@abhaygargab I have already implemented FairMOT. Just wanted to implement this MCMOT on my custom dataset. Thanks for letting me know these steps.
@abhaygargab I think you can figure out what happened by take a look at the results.txt output by the video to check detection results.
Dear @CaptainEven ,
python train.py --task mot --exp_id cater --gpus 1 --batch_size 8 --num_epochs 60 --lr_step '50' --data_cfg '../src/lib/cfg/cater.json' --load_model '../models/dla34-ba72cf86.pth' --output-root '../results' --arch dla_34
I used this command to train MCMOT for a task. I was able to successfully train the model for 60 epochs but the output video doesnt do any annotation or tracking. It is same as the input video.
This is the annotation for 1 frame of a video in the dataset. 2 1 0.262500 0.346388 0.114882 0.153176 3 1 0.318750 0.602854 0.217099 0.289465 4 1 0.465625 0.385947 0.089549 0.159198 1 1 0.581250 0.562053 0.081316 0.108421 4 2 0.275000 0.419104 0.143403 0.254940 2 2 0.725000 0.438366 0.060644 0.080858 2 3 0.434375 0.300891 0.051663 0.068884 4 3 0.628125 0.290727 0.122046 0.216971 4 4 0.803125 0.455257 0.145570 0.258791 4 5 0.509375 0.268900 0.036590 0.065049
Can you please help me figure out the reason why nothing is detected?
ThankYou