3 ms·
ha! Interesting. Did you train model yourself or find some ready available? BTW we also have an API for car's make/model/years/color/angle recognition. We can
by kosche 7y ago
ha! Interesting. Did you train model yourself or find some ready available?
BTW we also have an API for car's make/model/years/color/angle recognition. We can share API key for that if you'd like: https://carnet.ai https://carnet.ai
- godfreyho 7y agoWe collected images, labeled the data and trained our model all on our own. We now have about 330k images (about 1800 models). We used resnet152 to train it.
- kosche 7y agoI see. Sounds pretty similar to what we do.
- godfreyho 7y agoWe are just CS students and started this just for learning and experiencing with hands on computer vision and AI. So things are not very mature. Like we haven’t make the inferencing on the end user devices. The recognition runs on the server instead. And we don’t have many images as you do. We just crawled images on the web. Also, it doesn’t work very well with cars that can’t be found at where we lived, because we dont even know those models exist lol
- kosche 7y agoWe had/have this problem too. If nn "sees" unknown model it tries to classify it as one of known models. To avoid this, we added 'Other/Other' bucket and put ~3k images of different cars/buses/trucks in there. Now unknown models in most cases are classified as 'Other/Other' label.