Does BYOM in TAO Toolkit support object detection models
No, BYOM only supports Classification and Unet model.
Does that mean we can only use in-house pre-trained object detection models(yolov4, deformed DET etc. present in the catalog) while training through TAO Toolkit and we use can’t other models like YoloV5/V7?
TAO supports YOLO_v3, YOLO_v4, YOLO_v4_tiny, DINO, etc. User can train from scratch or train with pretrained models which are available in ngc.
BYOM(Bring Your Own Model) is a Python-based package that converts any open-source ONNX model to a TAO-compatible model. It helps users convert 3rd-party onnx model to a .tltb model which is compatible with TAO. BYOM only supports Classification and Unet model.
TAO does not include YOLO_v5 or v7.
There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.