How to make photos recognizable for system?

For optimal image recognition it is important to take correct pictures and avoid a bad user experience.
You have to keep in mind that our image recognition technology is based on the texture (visual elements) image recognition. In order to recognize them despite having different points of view, changes in the lightning, occlusion, etc. Therefore, the highest quality scores are given to images that show a rich texture.

We show below, in a very visual way, on the one hand the prototype of appropriate images and on the other hand, the ones that do not fit in order you to take into account when preparing images to be recognized.

  • Objects that work well
  • Objects that do not work

Which objects work well?

All objects with a particular texture or brands work well with our technology, for example: Logos and trademarks, newspapers and magazines, CD / DVD and book covers, Posters, packaged goods, monuments and places...In general, the more singularly texturized the surface of the object is, the more reliable and resistant to reflections, occlusions, inclination and fuzziness during the recognition.

What objects do not work well?

Although we are working to improve our technology, certain types of objects do not work well with our solution. Objects that do not work well are non-standard objects (for example, solid color only), dynamic in nature (like animals) or have thin bodies (eg cables, pencils, etc.)