中文版

Vehicle brand & model classification based on RetinaNet & Stanford Car Dataset

1. Detect vehicle in the image and mark with box. (Work on 196 brands)
2. Show the brand of the vehicle with probability.
3. Time spent during the detection.

git clone https://github.com/joey66666/vehicle-classify.git

cd vehicle-classify/vehicle_UI

conda install tensorflow==1.14.0 opencv numpy matplotlib keras

pip install pyqt5 keras-retinanet

cd keras_retinanet

mkdir snapshots

wget https://github.com/joey66666/vehicle-classify/releases/download/0.01/converted_restnet50_model.h5

python vehicle_ui.py

• It may take serval seconds to run the program related to the hardware, please wait after starting.
• Trained Models are here

• keras-retinanet
• Pyqt5
• Opencv
• Tensorflow
• Matplotlib

基于 RetinaNet 和 Stanford Car Dataset 的车辆型号检测识别方案

1. 检测画面中车辆并使用方框标记（可识别196种型号）
2. 给出车辆型号与可信度
3. 显示识别耗时

git clone https://github.com/joey66666/vehicle-classify.git

cd vehicle-classify/vehicle_UI

conda install tensorflow==1.14.0 opencv numpy matplotlib keras

pip install pyqt5 keras-retinanet

cd keras_retinanet

mkdir snapshots

wget https://github.com/joey66666/vehicle-classify/releases/download/0.01/converted_restnet50_model.h5

python vehicle_ui_cn.py

1. 根据不同硬件情况，识别可能耗费数十秒钟，请耐心等待

2. 训练后的模型在这里

• keras-retinanet
• Pyqt5
• Opencv
• Tensorflow
• Matplotlib