Open GreNait opened 4 years ago
Implemented the possibility, to detect figures with a trained model. All not detected images willbe saved for later training.
Deadpool is hard to detect, my guess si, that it is based on the model structure itself. I am looking into a new CNN structure based on the article: https://missinglink.ai/guides/tensorflow/tensorflow-conv2d-layers-practical-guide/
I found teh detection use on live webcam images. The images have the wrong format. I followed the answer form this thread: https://stackoverflow.com/users/12355605/grenait
and not it works!