Summary and Conclusions

We introduced DyPERS, a 'Dynamic Personal Enhanced Reality System' which uses computer vision and augmented reality to autonomously provide media memories related to real-world objects via a wearable platform. It allows a user to collect audio-visual clips in a seamless way and to retrieve them for playback automatically and meaningfully. We have described the three main building blocks of the system, namely the wearable hardware and interface, the generic object recognition system and the audio-visual associative memory. In addition, several application examples that DyPERS could span were enumerated. Experiments in a visual arts gallery environment suggest that the subjects using DyPERS would benefit of higher accuracy and more complete responses than participants using paper notes or no tools for information retention. These preliminary results are encouraging although more work is being planned to establish a final usability and performance evaluation. Nevertheless, the platform does provide interesting arguments for ways augmented reality and artificial perception can enrich the user and play a fundamental role in building a natural, seamless and intelligent interface.