README.md

PP-LiteSeg: A Superior Real-Time Semantic Segmentation Model

Input

(Image from https://www.cityscapes-dataset.com/downloads/)

• cityscapes: (1, 3, 512, 1024)
• camvid: (1, 3, 768, 1024)

Output

Usage

Automatically downloads the onnx and prototxt files on the first run. It is necessary to be connected to the Internet while downloading.

For the sample image,

$ python3 pp_liteseg.py

If you want to specify the input image, put the image path after the --input option.
You can use --savepath option to change the name of the output file to save.

$ python3 pp_liteseg.py --input IMAGE_PATH --savepath SAVE_IMAGE_PATH

By adding the --video option, you can input the video.
If you pass 0 as an argument to VIDEO_PATH, you can use the webcam input instead of the video file.

$ python3 pp_liteseg.py --video VIDEO_PATH

By adding the --model_type option, you can specify model type which is selected from "stdc1", "stdc2". (default is stdc1)

$ python3 pp_liteseg.py --model_type stdc1

By adding the --dataset option, you can specify model type (dataset) which is selected from "cityscapes", "camvid". (default is cityscapes)

$ python3 pp_liteseg.py --dataset cityscapes

Reference

• PaddleSeg
• PP-LiteSeg: A Superior Real-Time Semantic Segmentation Model

Framework

Pytorch

Model Format

ONNX opset=11

Netron

pp_liteseg_stdc1_cityscapes_1024x512_scale1.0.model.onnx.prototxt
pp_liteseg_stdc2_cityscapes_1024x512_scale1.0.model.onnx.prototxt
pp_liteseg_stdc1_camvid_960x720_10k_model.onnx.prototxt
pp_liteseg_stdc2_camvid_960x720_10k_model.onnx.prototxt