CrossTransformers: spatially-aware few-shot transfer

ankush-me / SynthText

Code for generating synthetic text images as described in "Synthetic Data for Text Localisation in Natural Images", Ankush Gupta, Andrea Vedaldi, Andrew Zisserman, CVPR 2016.

http://www.robots.ox.ac.uk/~vgg/data/scenetext/

Apache License 2.0

2.02k stars 622 forks source link

CrossTransformers: spatially-aware few-shot transfer #245

Closed Zhou-Yuen closed 2 years ago

Zhou-Yuen commented 2 years ago

Hello, I am not sure if you are the author of this paper, and I have some problems at present. I'm trying to visualize the softmax normalized attention weights ā as the Figure 3 in the CTX paper. I couldn't find the proper codes for this visualizing task anywhere in the repo. Can you please provide codes for this? Or at least give me some guideline for implementi

ankush-me commented 2 years ago

I indeed am. However would like to limit the discussion here to topics relevant to SynthText. Please email us your query —- email ids are given in the paper.