keras-yolo3

This is a copy from https://github.com/allanzelener/YAD2K.

Introduction

A Keras implementation of YOLOv3 (Tensorflow backend) inspired by allanzelener/YAD2K.

---

Quick Start

1. Download YOLOv3 weights from YOLO website.
2. Convert the Darknet YOLO model to a Keras model.

wget https://pjreddie.com/media/files/yolov3.weights
python convert.py yolov3.cfg yolov3.weights model_data/yolo.h5

3. Install dependencies

pip install -r requirements.txt

4. Put the directory with images in datasets.

5. Run YOLO to generate JSONs, and specify the directory path as bellow

python generate_json.py --dir_path datasets/test2017