zhuyingSeu
zhuyingSeu
For other models,such as mobilefacenet, resnet101, densenet121,how to get the compress rate?
We have trained ResNet series. You can follow our method train you model.
If you read the train.py, you will find this line :parser.add_argument("--patch_info", type=str, default="1_80x80", help="[org_1_80x60 / 1_80x80 / 2.7_80x80 / 4_80x80]")
If you have one gpu, python train.py --device_ids 0.
You can follow our method to train your model.
change the cv2.imread mode (src/data_io/dataset_folder.py, cv2.imread(image_path, 0)).
You can convert Pytorch model to TFlite model.
注意一下数据预处理，红外数据的预处理与RGB数据的预处理是不一样的。
> Where can I get Cap4M? Did you get Cap4M?
> > I tried setting the backbone to use FP16, the encoder-decoder part to use FP32. The results roughly match with the FP32 engine. But it is not as fast...