# FusionCount: Efficient Crowd Counting via Multiscale Feature Fusion

arXiv: https://arxiv.org/abs/2202.13660 · HTML: https://mlsanity.com/paper/2202.13660/

## Key facts

- Cited by 13 papers in the map (data through 14 Oct 2026).
- Builds on 11 of its 25 references that are in the map.
- First posted on arXiv 28 Feb 2022.

## Builds on (most cited first, 11 of 11)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [Very Deep Convolutional Networks for Large-Scale Image Recognition](https://mlsanity.com/paper/1409.1556/) (2014)
- [Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition](https://mlsanity.com/paper/1406.4729/) (2014)
- [CSRNet: Dilated Convolutional Neural Networks for Understanding the Highly Congested Scenes](https://mlsanity.com/paper/1802.10062/) (2018)
- [Bayesian Loss for Crowd Count Estimation with Point Supervision](https://mlsanity.com/paper/1908.03684/) (2019)
- [Context-Aware Crowd Counting](https://mlsanity.com/paper/1811.10452/) (2018)
- [Distribution Matching for Crowd Counting](https://mlsanity.com/paper/2009.13077/) (2020)
- [Generating High-Quality Crowd Density Maps using Contextual Pyramid CNNs](https://mlsanity.com/paper/1708.00953/) (2017)
- [TasselNet: Counting maize tassels in the wild via local counts regression network](https://mlsanity.com/paper/1707.02290/) (2017)
- [Encoder-Decoder Based Convolutional Neural Networks with Multi-Scale-Aware Modules for Crowd Counting](https://mlsanity.com/paper/2003.05586/) (2020)

## Built on by (most cited first, 13 of 13)

- [STEERER: Resolving Scale Variations for Counting and Localization via Selective Inheritance Learning](https://mlsanity.com/paper/2308.10468/) (2023)
- [Revisiting Crowd Counting: State-of-the-art, Trends, and Future Perspectives](https://mlsanity.com/paper/2209.07271/) (2022)
- [CLIP-EBC: CLIP Can Count Accurately through Enhanced Blockwise Classification](https://mlsanity.com/paper/2403.09281/) (2024)
- [CountFormer: Multi-View Crowd Counting Transformer](https://mlsanity.com/paper/2407.02047/) (2024)
- [Learning Discriminative Features for Crowd Counting](https://mlsanity.com/paper/2311.04509/) (2023)
- [FGENet: Fine-Grained Extraction Network for Congested Crowd Counting](https://mlsanity.com/paper/2401.01208/) (2024)
- [TreeFormer: a Semi-Supervised Transformer-based Framework for Tree Counting from a Single High Resolution Image](https://mlsanity.com/paper/2307.06118/) (2023)
- [Rethinking Cell Counting Methods: Decoupling Counting and Localization](https://mlsanity.com/paper/2503.13989/) (2025)
- [The Effectiveness of a Simplified Model Structure for Crowd Counting](https://mlsanity.com/paper/2404.07847/) (2024)
- [AdaTreeFormer: Few Shot Domain Adaptation for Tree Counting from a Single High-Resolution Image](https://mlsanity.com/paper/2402.02956/) (2024)
- [Evaluating Supervision Levels Trade-Offs for Infrared-Based People Counting](https://mlsanity.com/paper/2311.11974/) (2023)
- [SynMVCrowd: A Large Synthetic Benchmark for Multi-view Crowd Counting and Localization](https://mlsanity.com/paper/2603.23956/) (2026)
- [DCA-MoE: Spatially Adaptive Cross-Layer Fusion and Density-Routed Experts for Crowd Counting](https://mlsanity.com/paper/2608.15213/) (2026)

## Abstract (arXiv)

State-of-the-art crowd counting models follow an encoder-decoder approach. Images are first processed by the encoder to extract features. Then, to account for perspective distortion, the highest-level feature map is fed to extra components to extract multiscale features, which are the input to the decoder to generate crowd densities. However, in these methods, features extracted at earlier stages during encoding are underutilised, and the multiscale modules can only capture a limited range of receptive fields, albeit with considerable computational cost. This paper proposes a novel crowd counting architecture (FusionCount), which exploits the adaptive fusion of a large majority of encoded features instead of relying on additional extraction components to obtain multiscale features. Thus, it can cover a more extensive scope of receptive field sizes and lower the computational cost. We also introduce a new channel reduction block, which can extract saliency information during decoding and further enhance the model's performance. Experiments on two benchmark databases demonstrate that our model achieves state-of-the-art results with reduced computational complexity.