GesilaA

How to train the kp_detector model only?

3

The architecture of kp_detector defined in 'modules/keypoint_detector.py', and applied in 'GeneratorFullModel'. Loss function contains 'perceptual', 'generator_gan', 'feature_matching', 'equivariance_value', and 'equivariance_jacobian'. Which loss are designed for the kp_detector training？How to separate...

The model generate nothing after retrained on coco2014

1

Hi, you are handsome and your work are awesome! Thank you for your work. I got a problem when I try to train a smaller 'TransformerNet' model, all input images...

Failed to compile onnx model if use PSDK-RTOS's tidl_tools

1

SDK_version : ti-processor-sdk-rtos-j721s2-evm-08_05_00_11 TIDL_version : tidl_j721s2_08_05_00_16 I run commands below ``` export DEVICE=j7 export TIDL_TOOLS_PATH=${PSDKR_PATH}/tidl_j721s2_08_05_00_16/tidl_tools source ./setup.sh ``` and all passed. but error occured when I compile the onnx model:...

I have a puzzle about the tf.zeros_initializer() at line:85 in palm_detector.py ```conv_zeros = Conv2D(feature_n/2,2,strides=2,use_bias=False,kernel_initializer=tf.zeros_initializer())(tensor)``` Why this Conv2D operator should be initialized by tf.zeros_initializer()?

Thank you for you work! I clone this repository and download the pretrained weight from google drive. Run the command `python style.py ${content_image} ${style_image}`, The output of the model are...