MyPSP Robotics v4.4.4 – MyEye SIFT features for facial recognition testing

Homebrew developer ccpspita has pulled MyPSP Robotics version 4.4.3 up to version 4.4.4, successfully integrating the Scale Invariant Feature Transform (SIFT) algorithm for MyPSP’s MyEye module. The OpenCV powered module also makes use of PCA (Principal Components Analysis) to generate a feature selection based on a reference image and identify similar features on a target image. More to come for the Sony PSP robot project when you course the full story.

Download: MyPSP Robotics v4.4.4 – MyEye SIFT testbed

The latest update to roll in for the MyPSP Robotics project is a testbed for demonstrating how the MyEye facial recognition features work. PSP homebrew developer ccpspita integrated support for the SIFT (Scale Invariant Feature Transform) algorithm in MyPSP Robotics version 4.4.4, and the feature can be tested by following a two-step procedure for feature matching.

The latest build now includes a workspace for SIFT, where data used by the algorithm is generated and read from. That particular directory lies under the WORKAREA subdirectory of MyPSP Robotics main directory, aptly named SIFT. And putting that workspace to good use is a piece of cake, if you follow ccpspita’s instructions to the dot.

But first you’re going to have to download the 8MB archive and get it working the old fashioned way. All PRX plugins go to MyPSP Robotics’ PRX folder, and you’ll find the culprits semawm.prx, usbacc.prx, usbcam.prx, usbgps.prx, usbmic.prx, usbstor.prx, usbstorboot.prx, usbstormgr.prx, and usbstorms.prx hiding in flash0 – where they’d usually end up in.

When installation’s complete, take care to first read through the bundled documentation in this particular order: Readme, Message on OpenMIND, Message on MyEye OpenCV PCA (Principal Components Analysis), and Message on MyEye OpenCV SIFT. There are some training procedures you’ll have to run the application through first.

We’re not sure if it’s a particular requirement, but we’ve taken these steps to practice our newbie testing skills, and the SIFT testing procedure ran without a hitch. We could be wrong. If you’re still unsure, visit the release thread of MyPSP Robotics version 4.4.4 on QJ.NET’s forums to make the proper  inquiries.

Download: MyPSP Robotics v4.4.4 – MyEye SIFT testbed
Visit: QJ.NET’s PSP Development Forum