Pith. sign in

REVIEW 1 cited by

RepMLP: Re-parameterizing Convolutions into Fully-connected Layers for Image Recognition

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2105.01883 v3 pith:Y6Q7Z5RL submitted 2021-05-05 cs.CV cs.AIcs.LG

RepMLP: Re-parameterizing Convolutions into Fully-connected Layers for Image Recognition

classification cs.CV cs.AIcs.LG
keywords recognitionlayersrepmlpimagelocalpositionalconvolutionalface
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

We propose RepMLP, a multi-layer-perceptron-style neural network building block for image recognition, which is composed of a series of fully-connected (FC) layers. Compared to convolutional layers, FC layers are more efficient, better at modeling the long-range dependencies and positional patterns, but worse at capturing the local structures, hence usually less favored for image recognition. We propose a structural re-parameterization technique that adds local prior into an FC to make it powerful for image recognition. Specifically, we construct convolutional layers inside a RepMLP during training and merge them into the FC for inference. On CIFAR, a simple pure-MLP model shows performance very close to CNN. By inserting RepMLP in traditional CNN, we improve ResNets by 1.8% accuracy on ImageNet, 2.9% for face recognition, and 2.3% mIoU on Cityscapes with lower FLOPs. Our intriguing findings highlight that combining the global representational capacity and positional perception of FC with the local prior of convolution can improve the performance of neural network with faster speed on both the tasks with translation invariance (e.g., semantic segmentation) and those with aligned images and positional patterns (e.g., face recognition). The code and models are available at https://github.com/DingXiaoH/RepMLP.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. FW-GAN: Frequency-Driven Handwriting Synthesis with Wave-Modulated MLP Generator

    cs.CV 2025-08 conditional novelty 5.0

    FW-GAN combines wave-modulated MLP generation with frequency-domain losses and a wavelet-based discriminator to synthesize handwriting from one example, reporting lower FID scores than prior methods on IAM and HANDS-VNOnDB.