This is an implemetation of the VADNN article. This collection of scripts for Torch7 can compute both bottom-up and top-down visual attention saliency maps.
You need to fetch the trained model, at least, to be able to run the live demo.
./getData-public.shIf you want to use an USB camera, you will need to install the camera package
luarocks install cameraThere are available three different methods with which the final top-down saliency map is displayed.
Each method can be selected by --mode #, where # can be 1, 2 or 3.
The source can be a USB webcam, which is the default option, or an Ethernet camera, which can be chosen with --eth.
For example, we can run on USB, mode 3 by typing
cd dev
qlua saliencyCam.lua --mode 3or, we could run on Ethernet and using method 2 with
qlua saliencyCam.lua --mode 2 --eth