How to prune general yolov2 and yolov3 and yolov3 tiny models and use it for deepstream

Hi How to prune general yolov2 and yolov3 and yolov3 tiny models and use it for deepstream

For pruning, see https://docs.nvidia.com/metropolis/TLT/tlt-getting-started-guide/index.html#pruning_models

For deploy in DS, refer to https://docs.nvidia.com/metropolis/TLT/tlt-getting-started-guide/index.html#intg_yolov3_model

I had gone through the link. Can we directly prune the yolo darknet model and caffemodel without much steps

TLT can only prun the tlt format model which is trained by TLT.
For command, see the jupyter docker for reference too.

tlt-prune -m unpruned.tlt
-o pruned.tlt
-eq intersection
-pth 0.5
-k $KEY

can we convert a normal darknet model to tlt?

Please download
https://ngc.nvidia.com/catalog/models/nvidia:tlt_pretrained_object_detection/files?version=darknet53 or
https://ngc.nvidia.com/catalog/models/nvidia:tlt_pretrained_object_detection/files?version=darknet19
as a pretrained model and train with your own data.
The link a hdf5 file. After training, you will get a tlt model.

Oh so I can not use my own darknet weights and cfg file to prune…is there any other alternatives?

Yes, the own darknet is not compatible with tlt-prune.

Any otherways to prune general yoloV2 model and yoloV3 models so that I can use Deepstream

For tlt-prune, it can only prune tlt format model.