Retrain the Pretrained Model in Deepstream

AI & Data Science Other Products (closed) Miscellaneous Products (archived)
,
1

I want to retrain the Pre-trained object detection model which is provided in the deepstream.
The path to the model is as follows:

/opt/nvidia/deepstream/deepstream-5.1/samples/models/Primary_Detector

The files it contain are as follows:
cal_trt.bin, labels.txt, resnet10.caffemodel, resnet10.caffemodel_b1_gpu0_fp16.engine, resnet.prototxt

I want to retrain my model with my custom images.
Please guide me the steps for the same.

Thanks

2

You may find NVIDIA TAO can help you. It was previously called the NVIDA transfer learning toolkit.
https://developer.nvidia.com/tao
At the bottom of the page there are some resources with some examples. I found the GTC session quite helpful:

You may need to login using your Forum credentials (i.e yoiur NVIDIA Account) to get access.