search

ayooshkathuria / pytorch-yolo-v3

A PyTorch implementation of the YOLO v3 object detection algorithm
3.3k stars 1.06k forks source link

Is there any plan to add the training part? #61

Open Sephirot1st opened 5 years ago

TimDNN commented 5 years ago

That would be very helpful =)

BryanBao commented 5 years ago

https://pjreddie.com/darknet/yolo/

apple1987 commented 5 years ago

its mean, can i train by darknet，and using by the code?

fourth-archive commented 5 years ago

I've successfully trained YOLOv3 with PyTorch from scratch using this repo. It also does automatic image augmentation, and approaches the darknet mAP after about 65 epochs. https://github.com/ultralytics/yolov3

emmbertelen commented 5 years ago

I've successfully trained YOLOv3 with PyTorch from scratch using this repo. It also does automatic image augmentation, and approaches the darknet mAP after about 65 epochs. https://github.com/ultralytics/yolov3

How did you do it? That is what I exactly want. I want to use the YOLOv3 with PyTorch from scratch as the core of the code with the parts from the directory you linked which are necessary for training. Im not sure which parts should I copy besides the training.py, and which parts should I overwrite with in the YOLO from scratch with code from the ultralytics directory. Thanks for the help in advance.

glenn-jocher commented 5 years ago

@emmbertelen starting training is as easy as this in https://github.com/ultralytics/yolov3. You can even run this directly in our Colab Notebook: https://github.com/ultralytics/yolov3/blob/master/examples.ipynb

git clone https://github.com/ultralytics/yolov3  # clone
bash yolov3/data/get_coco_dataset_gdrive.sh  # copy COCO2014 dataset (20GB)
cd yolov3
python3 train.py  # train :)