README.md

Yolov5-Service

Yolov5-Service is a service that provides a RESTful API for Yolov5 models. run YOLOv5 with GPU. Easy As ABC

Inferencing

How to run

Install Requirements

• Docker (https://docs.docker.com/engine/install/
• Docker Compose (https://docs.docker.com/compose/install/)
• Docker nvidia engine (https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html#install-guide)

Put weights

Put your pretrained weights in the folder (for example, weights)

Docker

Config docker compose file (docker-compose.yml)

docker compose up

How to use

send sample image with test_req.py code.

Training

Send Train Request

{
  "label": "label Path or Coco annotation json",
  "image_path": "images path",
  "weights_path": "path tp pretrained weights",
  "is_augment": "true or false for data augmentation",
  "augment_params" : "Augmentation Params", -required if is_augment is True
  "validation_split": "Validation Split Number to split train and validation",
  "data_type": "Coco or yolo data type",  
  "image_size": "image size to resize",
  "log_url": "log url to get training status",
  "label_path": "label path to get label (used for yolo data type)",
  "save_dir": "save directory to save trained model"
}