search

paarthneekhara / text-to-image

Text to image synthesis using thought vectors
MIT License
2.16k stars 394 forks source link

Evaluation Metrics #3

Closed chingyaoc closed 8 years ago

chingyaoc commented 8 years ago

Hi, First of all, thanks for your awesome work! I'm wondering that is there any evaluation metrics for this kind of generative model since I want to compare the performance between using Skip Thought Vectors and the other embedding options for captions.

paarthneekhara commented 8 years ago

Hi, Thanks for your appreciation! As per the paper, I did not see any metric for evaluation of the generated images. There is only a qualitative comparison between the various generative models, by generating images from unseen captions. If you want to go ahead comparing different embedding options for captions, I strongly recommend also trying training them along with the model (this is suggested in the paper). A qualitative comparison can be made between different caption embeddings, as done between different generative models in the paper.

chingyaoc commented 8 years ago

thanks for your reply. Got the point!