How to train my own dataset

[InstantWindy]

Hi！ If I want to tain my own data with your model, how to run it? Thanks!

[glenn-jocher]

https://github.com/ultralytics/yolov3/wiki/Train-Custom-Data

[juanmed]

https://github.com/ultralytics/yolov3/wiki/Train-Custom-Data

Does the train.py file in the repository you link to works for with the darknet.py file in this (ayooshkathuria/pytorch-yolo-v3) repository?

[emmbertelen]

https://github.com/ultralytics/yolov3/wiki/Train-Custom-Data

Does the train.py file in the repository you link to works for with the darknet.py file in this (ayooshkathuria/pytorch-yolo-v3) repository?

Hey, did u find it out?

[glenn-jocher]

The YOLOv3 repository below is MacOS/Windows/Linux compatible, includes multigpu/multiscale, performs rectangular+batchnorm fused inference on images, videos, webcams, and even an iOS app. It tests slightly better than darknet (60.7 COCO2014 mAP with yolov3-spp.weights), and trains FP16/FP32 custom datasets from scratch to darknet performance, all using PyTorch :) https://github.com/ultralytics/yolov3

[FeatheryW]

@glenn-jocher I'm using a MacOS and have just finished training my dataset. How would I go by detecting my custom object using my webcam and or iphone? I have followed the tutorial you linked above.

[glenn-jocher]

Run this to use webcam:

git clone https://github.com/ultralytics/yolov3
cd yolov3
python3 detect.py --webcam

[NOxDriver]

@glenn-jocher Is there anyway to then use a custom trained weights on ios?

[glenn-jocher]

@NOssendryver of course. That’s the entire point of what we do. See YOLOv5 to get started.

[NOxDriver]

@glenn-jocher thanks. I've got my trained weights, but I just can not get it to work. What's the best way to get this done? Weights are in Darknet format. I can convert to Modelml, but nothing gets detected in the xcode project when I deploy to my iphone.

[glenn-jocher]

See YOLOV5 export tutorial https://github.com/ultralytics/yolov5#tutorials