Shazam for Fish !
One out of every three fish is mislabelled according to Oceana. Identifying fish in fillet form can be incredibly difficult to the naked eye. Oceana organizes an annual dinner with reknown sushi chefs each year. Despite their experience, they are consistently fooled. Instead of regular vision, we use infrared spectroscopy. This is a common scientifc technique that has been successfully applied to fish fillet authentication in this study.
Unfortunately, the study was limited in scope. No database of fish spectograms seem to exist. With funding, we hope to create a public one that catalogues the approximately 100 common species of seafood. We will be building upon the database by FishWatch.
We have built a DIY spectrometer upon the work of SpectralWorkbench. You can test our general demo here or our simple demohere which just calls your camera on your mobile phone. But, of course, with funding, we hope to use professional spectrometerswhich are more reliable. And thanks to quantum dot technology, spectrometers are becoming smaller and more portable.
We have built upon the audio fingerprintalgorithm which is the basis of Shazam. In fact, Shazam uses spectograms as well. It is just not recorded via infrared. Therefore, the algorithms are similar. We have extended the DejaVu open source projectwhich uses audio fingerprinting algorithms with the python language and numpylibrary.
Beyond our initial funding stage, we hope to be self-sustaining by providing a market referral service for reliable honest fish providers. In much the same way as Shazam provides links to download and buy songs, we will provide links to buy fish from verified wholesalers.