Nanodesktop Blind Assistant CFW0002

Developer Pegasus2000 visited the QJ.NET Forums earlier to announce the release of Nanodesktop Blind Assistant CFW0002. This homebrew’s updated build will carry improved algorithms for its facial recognition feature, as well as a new room recognition function. The rest of the changelog is available in the full article.

Download: Blind Assistant CFW0002
Visit: QJ.NET PSP Development Forum

Back in January, we broke the news regarding Pegasus2000‘s release of the Blind Assistant homebrew application for the Sony PSP. The application was interesting in that it was meant to be a tool for the blind, offering functions such as a facial recognition feature which enabled it to ID anybody whose face is recorded in its database.

The developer passed by the QJ.NET Forums earlier to announce that a new build for this application is ready for download. The face recognition algorithm was improved, and a new room recognition feature has been added. Quoting Pegasus2000’s list of changes for CFW0002:

• Blind Assistant has a new algorithm for face recognition. It has been hardly optimized for using Sony VFPU coprocessor. The speed is very high and recognition rate has been improved.
• Voice Synthesizer also uses VFPU . This allows to the software a generation of vocal samples in real time.
• Now, Blind Assistant is able also to recognize in which room the blind is. This task is done through a client/server architecture, that uses a new component called BlindServer. 
• Blind Assistant CFW0002 supports also a complex system for optical char recognition. The system is able to read a text or a label, through the Sony Go-Cam.

Pegasus2000 further explained that it’s not actually the PSP doing the calculations for the room recognition feature. Blind Assistant will instead use the PSP’s Wi-Fi connection to send a data packet to BlindServer, which will in turn use the SIFT (Scalar Invariant Feature Transform) algorithm to ID which room the unit is in.

After the data’s been processed, BlindServer sends it back to Blind Assistant. The developer explained this was necessary since the PSP’s CPU power would not be able to handle SIFT computations on its own, with BlindServer running on a more powerful processor.

We further remind the readers that they will need a Sony Go!Cam to run Blind Assistant’s functions. Full instructions on installing and programming this homebrew are available in the file bundle’s readme. We advise the readers to also visit Pegasus2000’s release thread at the QJ.NET PSP Development Forum for full details regarding this homebrew.

Download: Blind Assistant CFW0002
Visit: QJ.NET PSP Development Forum