Cannot use GPU training about mask-rcnn_tf2.14.0 HOT 2 CLOSED

HI @windvibra, sorry for my late response. Can you check if TensorFlow is built using CUDA? You can verify if TF is using GPU via:

import tensorflow as tf
print(tf.test.is_gpu_available()) # Should print True if TF can detect GPU
print("Num GPUs Available: ", len(tf.config.list_physical_devices('GPU'))) # Should print the number of GPUs available

Additionally, when you trained the model, did you check if the GPU memory was allocated to the model?

from mask-rcnn_tf2.14.0.

I may have an idea. TensorFlow has stopped GPU support from TF 2.11 on native-Windows platform. More details can be found here.

You'd have to either use Linux or WSL 2 on Windows to use this repo and TF with GPU support.

from mask-rcnn_tf2.14.0.