HeinSight3.0

code and models:

dataset:

Installation

Clone yolov5

git clone https://github.com/ultralytics/yolov5

Install python packages via commands:

conda create -n hte python=3.9
conda activate hte
cd yolov5
pip3 install -r requirements.txt
cd ..
pip3 install -r requirements.txt

Usage (images/videos):

Place the image/video in ./data/test_images. Then from root run

python3 main.py --input_image_path <path of video> --nms_iou 0.1 --conf 0.2 --batch_size 64 --create_plots

Example

python3 main.py --input_image_path ./data/test_images/hetro_homo_cap_flash.mp4 --nms_iou 0.1 --conf 0.2 --batch_size 64 --create_plots

Usage (camera):

From root run

python3 main.py --use_cameras <num_cameras> --nms_iou 0.1 --conf 0.2 --batch_size 64 --create_plots

<num_cameras> represents the number of cameras that you want to analyze

Example If you want to analyze 2 cameras

python3 main.py --use_cameras 2 --nms_iou 0.1 --conf 0.2 --batch_size 64 --create_plots

To finish camera detections gracefully, press and hold the esc key to stop the analysis

Publication:

https://chemrxiv.org/engage/chemrxiv/article-details/65e5481f9138d231619c1879