Pretrained model efficiency

DeepRNN / image_captioning

Tensorflow implementation of "Show, Attend and Tell: Neural Image Caption Generation with Visual Attention"

MIT License

785 stars 354 forks source link

Pretrained model efficiency #62

Closed zbeedatm closed 5 years ago

zbeedatm commented 5 years ago

I had used the pretrained model that you shared and I got bad results of captioning... Which cnn did you use for it? the default value from config.py which is vgg16? Can you share a model which was trained with resnet, and with more advanced configurations that can help getting better results?

In general, which configuration were you changing to achieve that?

Thanks in advance.