CS:GO Object Detection using YOLOv5

Clone yolov5 project and install required packages

git clone https://github.com/ultralytics/yolov5
cd yolov5
pip install -r requirements.txt

Dataset

Feel free to explore it here: https://drive.google.com/drive/folders/1qsz45_LiRjRpqTOBSIbX8WMv_nyb7gRJ

Manually upload it or use the following command to download it:

wget https://drive.google.com/file/d/1bqLrs6bXsBlDR19qPI0EaLsdPxG8gSyV/view?usp=sharing
unzip csgo_data.zip

Configuration

If you want to use a different dataset

• Modify the csgo.yaml file
• Modify yolov5s-c.yaml file (only change nc)

Project Files

• copy cfg/yolov5s-c.yaml to yolov5/models/ directory
• copy cfg/csgo.yaml to yolov5/data/ directory
• copy test.py to yolov5 main directory

Train

To train the model on CS:GO dataset, run the following command:

python train.py --img 640 --batch 16 --epochs 600 --data data/csgo.yaml --weights '' --cfg yolov5s-c.yaml

Feel free to change -epochs, and -batch size, if you want to use a pretrained model, change --weights.

Test

test.py is a custom implemented detection script to show IoU on test images, you can use YoloV5's test methods if you do not need IoU.

Setup

• Create a directory to save test output to (For an example test_out in yolov5 directory)
• Copy cs_weights.pt to main directory of yolov5 (Ignore this if you want to use your trained model)

Run

To test the model on CS:GO dataset, run the following command:

python test.py --data data/csgo.yaml --weights cs_weights.pt --dist test_out/

Output Example