Mask detection & social distancing detection using YOLOv5 with Intel Realsense D415/D435

Running Env Setup

Python 3.7 or later with all requirements.txt dependencies installed, including torch>=1.7. To install run:

pip install -r requirements.txt

Training @ Virtual Machine

A trained model has saved in /weights/best.pt

python train.py --batch 1 --epochs 200 --data ./data/data.yaml --cfg models/yolov5s.yaml --weights '' --device 0

Inference

• detect.py runs inference on a variety of sources, downloading models automatically from the latest YOLOv5 release and saving results to runs/detect.

$ python detect.py --source 0  # webcam
                            file.jpg  # image 
                            file.mp4  # video
                            path/  # directory
                            path/*.jpg  # glob
                            rtsp://170.93.143.139/rtplive/470011e600ef003a004ee33696235daa  # rtsp stream
                            rtmp://192.168.1.105/live/test  # rtmp stream
                            http://112.50.243.8/PLTV/88888888/224/3221225900/1.m3u8  # http stream

To run inference on example video in data/street.mp4:

python detect.py --source data/street.mp4 --weights weights/best.pt

To run inference on intel realsense camera:

python detect.py --source intel --weights weights/best.pt

Training

Run commands below to reproduce results on COCO dataset (dataset auto-downloads on first use). Training times for YOLOv5s/m/l/x are 2/4/6/8 days on a single V100 (multi-GPU times faster). Use the largest --batch-size your GPU allows (batch sizes shown for 16 GB devices).

$ python train.py --data coco.yaml --cfg yolov5s.yaml --weights '' --batch-size 64
                                         yolov5m                                40
                                         yolov5l                                24
                                         yolov5x                                16

Thanks to repos from ultralytics and iAmEthanMai, more ref