Building Docker image

sudo docker build -t  saytosid/segnet_demo:latest .

Running a container with this image

docker run -it --device /dev/video0 -v <ABSOLUTE PATH TO YOUR OUTPUT DIRECTORY>:/workspace/out -v <ABSOLUTE PATH TO YOUR INPUR DIRECTORY>:/workspace/in saytosid/segnet_demo

Running segnet

./segnet.sh  

This will run segnet on .jpg images of your <INPUT_DIRECTORY> and produce outputs in your <OUTPUT_DIRECTORY>

Demo

1.

2.

3.

Note:

The code is present in segnet_demo.py and is quite straight forward. If you change the code then you need to rebuild the docker image.

Based on

SegNet: A Deep Convolutional Encoder-Decoder Architecture for Robust Semantic Pixel-Wise Labelling 

For more details on the project please refer to the Segnet website
The docker image is based on aospan's docker-segnet image