search

alexander-pv / maskrcnn_tf2

Mask R-CNN for object detection and instance segmentation with Keras and TensorFlow V2 and ONNX and TensorRT optimization support.
Other
40 stars 11 forks source link

Regarding inference #10

Closed VeeranjaneyuluToka closed 2 years ago

VeeranjaneyuluToka commented 2 years ago

Hi,

I have noticed that you explicitly transferring the weights from training graph to inference graph. Would you please let me know the reason for the same?

Thanks, Veeru.

alexander-pv commented 2 years ago

Hi, @VeeranjaneyuluToka,

Basically, training and inference graphs differ for MaskRCNN. The original matterport model has the same feature with a bit less verbosity. Besides, such a way helps to port the model to .onnx.

You can consider these basic schemes for training and inference mode: mask_rcnn_training_mode

mask_rcnn_inference_mode