Padenti Documentation

Results for hand-parts labeling and class labelling on the MSRC2 and NYU2 datasets

(MSRC2,NYU2)

Padenti is an Open Source implementation of the Random Forests classifier specifically suited for Computer Vision applications that use simple per-pixel local features (e.g. class labeling, objects segmentation etc.). Both the training and the prediction are accelerated on GPUs using the OpenCL framework.

The library has been developed by the Engineering for Health and Wellbeing group at the Institute of Electronics, Computer and Telecommunication Engineering of the National Research Council of Italy (CNR).

Features include:

• fast training of large datasets using OpenCL
• support for both NVIDIA and AMD GPUs
• support of arbitrary image pixel type and number of channels
• support of arbitrary per-pixel features through a custom OpenCL C function

"Padenti" stands for "Forest" in Sardinian language (in its variant of the Mogoro village).

For installation, usage and a small tutorial please consult the following sections:

• Installation Installation and Usage
• Tutorial Tutorial