1. How to compare 2 numpy arrays online

There are no reliable online tools. 2 options. One is in python, other is in pycharm. I wanna explain about the latter. Load the numpy array using np.load and then hit the debug mode in pycharm. View the variable you loaded as array in pycharm and then export it to csv files. Then use any online csv compare tool to compare the arrays

2. View the inputs and outputs of a frozen_graph

  • Replace graph_inputs by graph_outputs to print the outputs
import graphsurgeon as gs
graph = gs.DynamicGraph('./mask_rcnn_inception_v2_coco_2018_01_28/frozen_inference_graph.pb')

print(graph.graph_inputs)

Make sure to install pip install graphsurgeon

3. Freeze all the packages in your pip environment to the requirements.txt ➡️ pip3 freeze > requirements.txt

4. Adding frame rate or fps to the opencv frames

  • https://learnopencv.com/how-to-find-frame-rate-or-frames-per-second-fps-in-opencv-python-cpp/

5. If CUDA goes out of memory while running on pycharm

import os
os.environ['TF_FORCE_GPU_ALLOW_GROWTH'] = 'true'

6. To see if the GPU is available for training in tensorflow 

print("Num GPUs Available: ", len(tf.config.experimental.list_physical_devices('GPU')))

7. Tensorboard basics 

tensorboard --logdir logs/fit

Tunnelling tensorboard

#from your local machine, run

ssh -N -f -L localhost:16006:localhost:6006 <user@remote>

#on the remote machine, run:

tensorboard --logdir <path> --port 6006