# WeightNet: Revisiting the Design Space of Weight Networks

arXiv: https://arxiv.org/abs/2007.11823 · HTML: https://mlsanity.com/paper/2007.11823/

## Key facts

- Cited by 50 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 23 Jul 2020.

## Built on by (most cited first, 30 of 50)

- [Dynamic Neural Networks: A Survey](https://mlsanity.com/paper/2102.04906/) (2021)
- [Mobile-Former: Bridging MobileNet and Transformer](https://mlsanity.com/paper/2108.05895/) (2021)
- [Dynamic Spatial Propagation Network for Depth Completion](https://mlsanity.com/paper/2202.09769/) (2022)
- [Omni-Dimensional Dynamic Convolution](https://mlsanity.com/paper/2209.07947/) (2022)
- [Decoupled Dynamic Filter Networks](https://mlsanity.com/paper/2104.14107/) (2021)
- [Adaptive Rotated Convolution for Rotated Object Detection](https://mlsanity.com/paper/2303.07820/) (2023)
- [Revisiting Dynamic Convolution via Matrix Decomposition](https://mlsanity.com/paper/2103.08756/) (2021)
- [Reversible Column Networks](https://mlsanity.com/paper/2212.11696/) (2022)
- [PointConvFormer: Revenge of the Point-based Convolution](https://mlsanity.com/paper/2208.02879/) (2022)
- [MicroNet: Improving Image Recognition with Extremely Low FLOPs](https://mlsanity.com/paper/2108.05894/) (2021)
- [Residual Quantization with Implicit Neural Codebooks](https://mlsanity.com/paper/2401.14732/) (2024)
- [DyFADet: Dynamic Feature Aggregation for Temporal Action Detection](https://mlsanity.com/paper/2407.03197/) (2024)
- [Frequency Dynamic Convolution for Dense Image Prediction](https://mlsanity.com/paper/2503.18783/) (2025)
- [ELSA: Enhanced Local Self-Attention for Vision Transformer](https://mlsanity.com/paper/2112.12786/) (2021)
- [Relational Self-Attention: What's Missing in Attention for Video Understanding](https://mlsanity.com/paper/2111.01673/) (2021)
- [AGConv: Adaptive Graph Convolution on 3D Point Clouds](https://mlsanity.com/paper/2206.04665/) (2022)
- [RRSR:Reciprocal Reference-based Image Super-Resolution with Progressive Feature Alignment and Selection](https://mlsanity.com/paper/2211.04203/) (2022)
- [Decomposed Guided Dynamic Filters for Efficient RGB-Guided Depth Completion](https://mlsanity.com/paper/2309.02043/) (2023)
- [Dynamic Fusion Network for RGBT Tracking](https://mlsanity.com/paper/2109.07662/) (2021)
- [RefConv: Re-parameterized Refocusing Convolution for Powerful ConvNets](https://mlsanity.com/paper/2310.10563/) (2023)
- [TVConv: Efficient Translation Variant Convolution for Layout-aware Visual Processing](https://mlsanity.com/paper/2203.10489/) (2022)
- [KernelWarehouse: Rethinking the Design of Dynamic Convolution](https://mlsanity.com/paper/2406.07879/) (2024)
- [Collaboration of Experts: Achieving 80% Top-1 Accuracy on ImageNet with 100M FLOPs](https://mlsanity.com/paper/2107.03815/) (2021)
- [PSLT: A Light-weight Vision Transformer with Ladder Self-Attention and Progressive Shift](https://mlsanity.com/paper/2304.03481/) (2023)
- [KernelWarehouse: Towards Parameter-Efficient Dynamic Convolution](https://mlsanity.com/paper/2308.08361/) (2023)
- [DRPN: Making CNN Dynamically Handle Scale Variation](https://mlsanity.com/paper/2112.10963/) (2021)
- [Encoding Semantic Priors into the Weights of Implicit Neural Representation](https://mlsanity.com/paper/2406.04178/) (2024)
- [3D Human Pose Lifting with Grid Convolution](https://mlsanity.com/paper/2302.08760/) (2023)
- [Dynamic Mobile-Former: Strengthening Dynamic Convolution with Attention and Residual Connection in Kernel Space](https://mlsanity.com/paper/2304.07254/) (2023)
- [Evolution: A Unified Formula for Feature Operators from a High-level Perspective](https://mlsanity.com/paper/2305.14409/) (2023)

## Abstract (arXiv)

We present a conceptually simple, flexible and effective framework for weight generating networks. Our approach is general that unifies two current distinct and extremely effective SENet and CondConv into the same framework on weight space. The method, called WeightNet, generalizes the two methods by simply adding one more grouped fully-connected layer to the attention activation layer. We use the WeightNet, composed entirely of (grouped) fully-connected layers, to directly output the convolutional weight. WeightNet is easy and memory-conserving to train, on the kernel space instead of the feature space. Because of the flexibility, our method outperforms existing approaches on both ImageNet and COCO detection tasks, achieving better Accuracy-FLOPs and Accuracy-Parameter trade-offs. The framework on the flexible weight space has the potential to further improve the performance. Code is available at https://github.com/megvii-model/WeightNet.