ObjectDetection with YOLOv2

You only look once (YOLO) is a state-of-the-art, real-time object detection system.

Usage

$ git clone https://github.com/thtrieu/darkflow.git

$ cd darkflow

$ wget https://pjreddie.com/media/files/yolov2-tiny.weights

$ wget https://github.com/pjreddie/darknet/blob/master/cfg/yolov2-tiny.cfg

$ flow --model cfg/tiny-yolo.cfg --load bin/tiny-yolo.weights

Change codes as your file paths on CFG_FILE_EXTENSION, WEIGHT_FILE_EXTENSION and IMAGE_PATH in objectDetection_yolo_v2

For example

options = {"model": "CFG_FILE_EXTENSION", "load": "WEIGHT_FILE_EXTENSION", "threshold": 0.1}

$ flow --model cfg/tiny-yolo.cfg --load bin/tiny-yolo.weights

After editing paths in python codes, you can run any python code in repo.

Tiny Yolo

• Dataset: COCO trainval
• mAp :23.7 Bn
• FLOPS :5.41Bn

Link for pretrained weight

https://pjreddie.com/media/files/yolov2-tiny.weights

Link for cfg file

https://github.com/pjreddie/darknet/blob/master/cfg/yolov2-tiny.cfg

Dependencies

tensorflow (1.10.0)