Why do we take the first channel feature map of image features when inferring？

zhjohnchan / R2GenCMN

[ACL-2021] The official implementation of Cross-modal Memory Networks for Radiology Report Generation.

Apache License 2.0

77 stars 7 forks source link

Closed Xjmengnieer closed 1 year ago

Xjmengnieer commented 1 year ago

As shown in the title, the code is as follows：

could you give me some advice ? Thanks

zhjohnchan commented 1 year ago

Hi @Xjmengnieer,

Thanks for your attention! It doesn't mean the first channel of the features. It just reshapes the tensors.

Best, Zhihong