archai/devices/vision
Chris Lovett f52e175aa2 fix: move to python 3.8 for device code. 2022-12-16 16:26:47 -03:00
..
collect_metrics.py fix: move to python 3.8 for device code. 2022-12-16 16:26:47 -03:00
readme.md Some readme improvements and move some snpe utility code. (#91) 2022-12-16 16:26:47 -03:00
screenshot.png initial commit of qualcomm device code 2022-12-16 16:26:46 -03:00

readme.md

Readme

  1. Visualize Mask R-CNN outputs and verify if the model outputs are correct. Inside your experiment folder, run collect_metrics.py --help to see the command line args. The --show option visualizes the results, for example:

    python collect_metrics.py --input d:\datasets\FaceSynthetics --output .\Results --show
    

    You should see something like this:

    screenshot

  2. Compute metrics. Inside your experiment folder, run collect_metrics.py again without the --show argument and you will get a .csv output file listing the F1 scores for each category and an overall score.