does generate prototxt only work for SSD, Retinanet and Yolov3, not for other custom models?

[fiefdx]

I saw the code, it looks like it only support SSD, Retinanet and Yolov3 models to generate prototxt, could we support other custom models, or maybe have some method to call.

[mathmanu]

Specifically which model are you interested in?

[mathmanu]

All the models supported by TIDL Detection Meta Arch Proto can be exported this way. So support for unsupported models can be added by looking at the models that are already supported in prototxt export.

See the detection meta archs supported by TIDL here: https://github.com/TexasInstruments/edgeai-mmdetection/blob/master/xmmdet/utils/proto/mmdet_meta_arch.proto

[mathmanu]

We also have support for

EfficientDetLite: https://github.com/TexasInstruments/edgeai-modelzoo/blob/master/models/vision/detection/coco/google-automl/README.md https://github.com/TexasInstruments/edgeai-modelzoo/tree/master/models/vision/detection (full support for EfficientDetLite is coming in the upcoming version of TIDL) Note: The detection head of EfficientDetLite is similar to RetinaNet

YOLOV5: https://github.com/TexasInstruments/edgeai-yolov5 (you can look at the prototxt examples provided in this repo and write the prototxt yourself for exported onnx - automatic export of prototxt is not supported in edgeai-yolov5)

YOLOX: Support is coming soon

In addition to edgeai-mmdetection, you can also train detection models using: https://github.com/TexasInstruments/edgeai-torchvision (currently only SSD prototxt export is enabled)

[zafeerali943]

@mathmanu brother can please tell how i can able to convert my yolov8 models to be embedded friendly that is especially for TDA4VM, kindly help me out.