Cant get to work facemak NVIDIA-AI-IOT/face-mask-detection

carlosdavid663 · November 16, 2020, 2:06pm

Hello everyone!!

Im starting on the jetson plattform and i want to try the [facemask detection from Nvidia](https://github.com/NVIDIA-AI-IOT/face-mask-detection#nvidia-transfer-learning-toolkit-tlt-training-flow-) but i find my self a bit lost here hopefully i can find some help here, I have listed pretty much every single problem that i encounter after training the data set, please ask me if i was not very clear or if you need to know more. Thanks!!!

i cant find this file( that i am aware i need to find it)

Transfer model files (.etlt), if int8: calibration file (calibration.bin)
Use config files from /ds_configs/* $vi config_infer_primary_masknet.txt
- Modify model and label paths: according to your directory locations

I need to modify this line ( tlt-encoded-model=/mnt/8c3f68c9-a08a-400b-8c80-99c5fee26a06/detectnet_v2_models/detectnet_4K-fddb-12/resnet18_RGB960_detector_fddb_12_int8.etlt ) with my directory location my directory location to that file is this (/home/nx/Downloads/face-mask-detection-master/ds_config ) how will it look like?

Look for tlt-encoded-model, labelfile-path, model-engine-file, int8-calib-file\

(where do i look for this information?? and how do i modify the tresholds?)
Modify confidence_threshold, class-attributes according to training

Look for classifier-threshold, class-attrs

i could not find this files as well and most likely i wont know how to modify them
se deepstream_config files: $ vi deepstream_app_source1_masknet.txt

Modify model file and config file paths:
- Look for model-engine-file, config-file under primary-gie

if it helps in anything here is my directory (/home/nx/Downloads/face-mask-detection-master)

this is the command i used to train

python3 data2kitti.py --kaggle-dataset-path “/home/nx/Downloads/face-mask-detection-master”
–mafa-dataset-path “/home/nx/Downloads/face-mask-detection-master/face-mask-detection-data/MAFA Dataset”
–fddb-dataset-path “/home/nx/Downloads/face-mask-detection-master”
–widerface-dataset-path “/home/nx/Downloads/face-mask-detection-master/face-mask-detection-data/Wider Face Dataset”
–kitti-base-path “/home/nx/Downloads/face-mask-detection-master/face-mask-detection”
–category-limit 6000
–tlt-input-dims_width 500 \ (i dont know why i used 500 but it worked if anyone could explain as well ill be very grateful for it)
–tlt-input-dims_height 500
–train (here should i put a number? )

the output of training

Kaggle Dataset: Total Mask faces: 4154 and No-Mask faces:790
Total Mask Labelled:4154 and No-Mask Labelled:790
Directory Already Exists
Directory Already Exists
/home/nx/Downloads/face-mask-detection-master/data_utils/mafa2kitti.py:51: RuntimeWarning: overflow encountered in ubyte_scalars
bbox = [_bbox_label[0], _bbox_label[1], _bbox_label[0]+_bbox_label[2], _bbox_label[1]+_bbox_label[3]]
MAFA Dataset: Total Mask faces: 6000 and No-Mask faces:232
Total Mask Labelled:6000 and No-Mask Labelled:232

Final: Total Mask Labelled:6000
Total No-Mask Labelled:4621

I

AastaLLL · November 17, 2020, 2:59am

Hi,

Is this issue duplicate to the Cant get to work facemak NVIDIA-AI-IOT/face-mask-detection or for other issues?

Thanks.

carlosdavid663 · November 18, 2020, 5:09am

Its a different one , i cant find the files on this one are yhey supposed to be generated downloaded ?

carlosdavid663 · November 18, 2020, 4:27pm

i still cant find where is it or how to do it. i have followed it each step and i still cant find those files.

carlosdavid663 · November 29, 2020, 6:44am

is any one available to help?

AastaLLL · November 30, 2020, 6:25am

Hi,

1..
Have you applied the steps listed below:

github.com

NVIDIA-AI-IOT/face-mask-detection/blob/master/face-mask-detection.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Face Mask Detection using NVIDIA TLT "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The MIT License (MIT)\n",
    "\n",
    "Copyright (c) 2019-2020, NVIDIA CORPORATION.\n",
    "\n",
    "Permission is hereby granted, free of charge, to any person obtaining a copy of\n",
    "this software and associated documentation files (the \"Software\"), to deal in\n",
    "the Software without restriction, including without limitation the rights to\n",

This file has been truncated. show original

The pre-trained model and calibration file can be downloaded from our NGC website or generated by training.
Please follow the steps above and you should be able to get them at step-1.

2.
To modified the path, please check the configure file below:
https://github.com/NVIDIA-AI-IOT/face-mask-detection/blob/master/ds_configs/config_infer_primary_masknet_gpu.txt
Then update the path based on your environment. For example:

tlt-encoded-model=/home/nvidia/resnet18_RGB960_detector_fddb_12_int8.etlt

3.
confidence_threshold is updated based on the training parameter you are used.
If a pre-trained model is used, you can leave it unchanged.

4.
All the configure file is located at {folder}/ds_configs/.

Thanks.

carlosdavid663 · December 6, 2020, 7:08pm

Yeah i still couldnt i guess this is for super advanced users , it would be nice a tutorial, thanks though

Cant get to work facemak NVIDIA-AI-IOT/face-mask-detection

Final: Total Mask Labelled:6000 Total No-Mask Labelled:4621

Final: Total Mask Labelled:6000
Total No-Mask Labelled:4621