Training PyTorch-YOLOv3 with multi-GPUs.
$ git clone https://github.com/zhangyongshun/Training_YOLOv3_with_Multi-GPUs.git
$ cd Training_YOLOv3_with_Multi-GPUs/
$ sudo pip install -r requirements.txt
$ cd weights/
$ bash download_weights.sh
$ cd data/
$ bash get_coco_dataset.sh
To train on COCO using a Darknet-53 backend pretrained on ImageNet run:
$ Modify the configs in config/config.py, especially paths of data and weights.
$ python train.py
We make the same number of targets for each image in function ListDataset.collate_fn of utils/dataset.py.