search

shepnerd / inpainting_gmcnn

Image Inpainting via Generative Multi-column Convolutional Neural Networks, NeurIPS2018
MIT License
428 stars 98 forks source link

Input to generator #10

Closed tlatkowski closed 5 years ago

tlatkowski commented 5 years ago

Hi @shepnerd ,

I've analysed your code and I have a question regarding the input to the generator. While making the input you concatenate the masked image with mask and matrix of ones. I understand pros of using mask as additional input but what is the intuition behind the usage of one more channel filled with 1?

shepnerd commented 5 years ago

That part can be removed if no more conditions are fed in. Besides of just inpainting a missing region, sometimes we can give more conditions to guide the generation process, e.g., semantic layout, edge maps. These types of additional conditions can be added as extra channels of the input. In this implementation, however, the aforementioned extra conditions are not utilized.

Tomasz Latkowski notifications@github.com 于2019年4月3日周三 下午5:52写道：

Hi @shepnerd https://github.com/shepnerd ,

I've analysed your code and I have a question regarding the input to the generator. While making the input you concatenate the masked image with mask and matrix of ones. I understand pros of using mask as additional input but what is the intuition behind the usage of one more channel filled with 1?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/shepnerd/inpainting_gmcnn/issues/10, or mute the thread https://github.com/notifications/unsubscribe-auth/AE1PCC4Lsz6wiODKg45kQNDgUHT8CQ4Eks5vdHnzgaJpZM4caF9B .

tlatkowski commented 5 years ago

Great thanks!