[Refactor] RefineNet: SEBlock, Residual Connection & Import Cleanup#134
Open
agentksimha wants to merge 28 commits intohumanai-foundation:mainfrom
Open
[Refactor] RefineNet: SEBlock, Residual Connection & Import Cleanup#134agentksimha wants to merge 28 commits intohumanai-foundation:mainfrom
agentksimha wants to merge 28 commits intohumanai-foundation:mainfrom
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What changed in code/CRAFT/refinenet.py:
SEBlock — Added channel-wise recalibration (AdaptiveAvgPool2d → Linear → ReLU → Linear → Sigmoid) applied after last_conv, before ASPP branches. Helps the network weight which feature channels matter most.
Residual connection — Added skip_proj = Conv2d(34→64, kernel_size=1) with refine = last_conv(x) + skip_proj(x). Improves gradient flow through the refinement block.
Removed Variable import — Was unused and deprecated since PyTorch 0.4.
No breaking changes. ASPP branches, output interface, and init_weights calls are untouched. Drop-in replacement.