keras-yolo3-kitti

base on keras-yolo3

Introduction

A Keras implementation of YOLOv3 worked on kitti dataset.

Training

1. Generate your own annotation file and class names file.
   One row for one image;
   Row format: image_file_path box1 box2 ... boxN;
   Box format: x_min,y_min,x_max,y_max,class_id (no space).
   For VOC dataset, try python voc_annotation.py
   Here is an example:

   path/to/img1.jpg 50,100,150,200,0 30,50,200,120,3
   path/to/img2.jpg 120,300,250,600,2
   ...

   There is a notebook jupyter called kitti_train.ipynb The kitti dataset is Kitti Object,I only use 3 classes.

2. Make sure you have run python convert.py -w yolov3.cfg yolov3.weights model_data/yolo_weights.h5
   The file model_data/yolo_weights.h5 is used to load pretrained weights.

3. Modify train.py and start training.
   python train.py
   Use your trained weights or checkpoint weights in yolo.py.
   Remember to modify class path or anchor path.

4 test the images python yolo.py -s test_images -d output_images Label images in folder test_images into folder output_images

Some issues to know

In convert.py I have some hard code for the pretrained model. You can also use the original code. I write a blog in Chinese. Here is the address. https://blog.csdn.net/yangchengtest/article/details/80732237