TanTechnova 2023

This is our submission to the TanTechnova 2023.

The project supports video dehazing, live on the Raspberry Pi 400.

Prototype

• Prototype is a blender file named 3D Prototype.blend, it can be imported in blender.

Run:

• The below is for Raspberry Pi implementation of full project.
• Make sure that the shell environment is in the directory with main.py, and dehazer.pth

#startup.sh
$ cd /path/to/python/folder
$ python3 main.py 

• To ensure the file opens at startup of desktop environment, place to add in other folders/startup.desktop in /etc/xdg/autostart/.

• Add the location of startup.sh in To add in other folders/startup.desktop.

Exec=/path/to/startup.sh

• Make sure to set startup.sh's permissions to rwxr-xr-x so it can be executed.

chmod rwxr-xr-x /path/to/startup.sh

• startup.sh is recommended to be placed in /path/to/python/file

• In startup.sh, make sure to change the directory path.

#!/bin/sh

cd /path/to/python/folder
python3 main.py

• /Others/Wallpaper.png has been set as the background image of the desktop environnment.

Train, test.:

See original repository for information on training.

• Train there, import the trained Model (dehazer.pth) here.
• This is because we don't focus on training, or the model. The pretrained one provided here is already good.