3rd of CVPPA@ECCV'2024: Detection and Multi-Object Tracking of Sweet Peppers Challenge

challenge link

official workshop

create environment

## 克隆仓库 clone repository
git clone git@github.com:chunbai1/ECCV-CVPPA-2024-MOT.git
cd ECCV-CVPPA-2024-MOT

## torch
conda create -n bytetrack python=3.7
pip install torch==1.9.1+cu111 torchvision==0.10.1+cu111 torchaudio==0.9.1 -f https://download.pytorch.org/whl/torch_stable.html

## other package 
pip3 install -r requirements.txt 
python3 setup.py develop
pip3 install cython; pip3 install 'git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI'
pip3 install cython_bbox

Dataset

First, download source dataset from official Link. Then follow the steps below to process the data into the MOT20 dataset format.

After downloading the official dataset and decompressing it, the format is as follows:

.MOT_ECCV_2024
  ├── depth
  ├── mask2former_output
  ├── rgb
  └── weak_labels

According to the official dataset description file, the following dataset format is obtained:

.
├── test
│   ├── depth
│   ├── mask2former_output
│   ├── mask2former_txt
│   ├── rgb
│   └── weak_labels
├── train
│   ├── depth
│   ├── mask2former_output
│   ├── rgb
│   └── weak_labels
└── valid
    ├── depth
    ├── mask2former_output
    ├── rgb
    └── weak_labels

Then set the --subset parameters to 'train', 'val', 'test' respectively, and execute the my_toos/convert2mot20.py file, convert the MOT20 data format to COCO format by my_tools/convert2coco.py. So the data format is converted to:

.
├── annotations
│   ├── test.json
│   ├── train.json
│   ├── train_half.json
│   └── val_half.json
├── test
├── train

Finally, create a dataset directory and establish a soft link according to the my_tools/mix_data.py file to get the final dataset.

Train

Change the exps/example/mot/yolox_x_mix_det.py configuration to get exps/example/mot/yolox_x_mix_det_cvppa.py file. You can refer ByteTrack_README.md.

CUDA_VISIBLE_DEVICES=0,1,2,3 python3 tools/train.py -f exps/example/mot/yolox_x_mix_det_cvppa.py -d 4 -b 16 --fp16 -o \
-c /data/ChaiJM/Competition/CVPPA-DMOT/Code/ByteTrack-main/YOLOX_outputs/yolox_x_mix_det_cvppa/latest_ckpt.pth.tar \
-expn cvppa_resume  -e 4

Test

CUDA_VISIBLE_DEVICES=7 python tools/demo_track_cvppa.py --save_result -expn cvppa-test5 \
-c pth

Post-Processing

Refer blog.

Refer

ByteTrack