Open 0-ali opened 5 years ago
Yes, that's definitely in my plans. Make an interface where the user can submit an image and the caption and the model gets re-trained accordingly. That will increase the usability of this by miles.
The model retraining is totally possible.
What is the method to retrain the model? Where to get the training script?
I was wondering if I can re-train
flickr8k_cnn_lstm_v1.p
model