Research map / Detection and segmentation / CNN features
Conv2Former: A Simple Transformer-Style ConvNet for Visual Recognition
Qibin Hou, Cheng-Ze Lu, Ming‐Ming Cheng, Jiashi Feng
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| increase depthwise kernel size from 5x5 to 21x21 (ablation) | Conv2Former-T · ImageNet-1k | top-1 accuracy: 82.8 -> 83.4 | Improved |
| increase depthwise kernel size from 5x5 to 21x21 (ablation) | Conv2Former-B · ImageNet-1k | top-1 accuracy: 84.1 -> 84.5 | Improved |
| use 11x11 kernel instead of 7x7 (ablation) | Conv2Former-B · ImageNet-1k | top-1 accuracy: 84.2 -> 84.4 | Improved |
| add Hadamard product modulation instead of element-wise sum (ablation) | Conv2Former-T · ImageNet-1k | top-1 accuracy: 82.7 -> 83.2 | Improved |
| add Sigmoid after the convolutional weights A (ablation) | Conv2Former-T · ImageNet-1k | top-1 accuracy: 83.2 -> 82.3 | Hurt |
| add L1 normalization after A (ablation) | Conv2Former-T · ImageNet-1k | top-1 accuracy: 83.2 -> 82.8 | Hurt |
| linearly normalize A to (0, 1] (ablation) | Conv2Former-T · ImageNet-1k | top-1 accuracy: 83.2 -> 82.2 | Hurt |
| use Layer Normalization with GELU (ablation) | Conv2Former · ImageNet-1k | top-1 accuracy: +0.1%-0.2% | Improved |
| use deeper stage configuration 3-3-12-3 instead of 2-2-8-2 for tiny models (ablation) | Conv2Former · ImageNet-1k | top-1 accuracy: 81.5 -> 83.2 | Improved |
| use three convolutions for patch embedding instead of one (ablation) | Conv2Former-IS · ImageNet-1k | top-1 accuracy: 81.2 -> 82.0 | Improved |
| use three convolutions for patch embedding instead of one (ablation) | Conv2Former-IB · ImageNet-1k | top-1 accuracy: 82.7 -> 83.0 | Improved |
| replace ConvNeXt block with convolutional modulation block | ConvNeXt-T · COCO Mask R-CNN | box AP: 46.2 -> 48.0 | Improved |
Closest papers by meaning
- A ConvNet for the 2020s (CVPR 2022 · similarity 0.92)
- Is Second-Order Information Helpful for Large-Scale Visual Recognition? (ICCV 2017 · similarity 0.90)
- CvT: Introducing Convolutions to Vision Transformers (ICCV 2021 · similarity 0.90)
- ConvNeXt V2: Co-Designing and Scaling ConvNets With Masked Autoencoders (CVPR 2023 · similarity 0.90)
- DECO: Unleashing the Potential of ConvNets for Query-based Detection and Segmentation (ICLR 2025 · similarity 0.89)
- UniRepLKNet: A Universal Perception Large-Kernel ConvNet for Audio Video Point Cloud Time-Series and Image Recognition (CVPR 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
