search

aimagelab / meshed-memory-transformer

Meshed-Memory Transformer for Image Captioning. CVPR 2020
BSD 3-Clause "New" or "Revised" License
517 stars 136 forks source link

Test on Custom Dataset #89

Closed Faiail closed 1 year ago

Faiail commented 1 year ago

Is there a possibility to test your model giving images as input? I'm struggling in extracting features, using bottom-up-attention repository. Thabks

marcellacornia commented 1 year ago

Hi @Faiail, thanks for your interest in our work! To extract features we used this repository. If you want to test our model on images outside the COCO dataset, you need to extract features using the same source code.