# Crowd counting via scale-adaptive convolutional neural network

arXiv: https://arxiv.org/abs/1711.04433 · HTML: https://mlsanity.com/paper/1711.04433/

## Key facts

- Cited by 23 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 13 Nov 2017.

## Built on by (most cited first, 23 of 23)

- [STEERER: Resolving Scale Variations for Counting and Localization via Selective Inheritance Learning](https://mlsanity.com/paper/2308.10468/) (2023)
- [Detection, Tracking, and Counting Meets Drones in Crowds: A Benchmark](https://mlsanity.com/paper/2105.02440/) (2021)
- [Domain-General Crowd Counting in Unseen Scenarios](https://mlsanity.com/paper/2212.02573/) (2022)
- [Direct Measure Matching for Crowd Counting](https://mlsanity.com/paper/2107.01558/) (2021)
- [Redesigning Multi-Scale Neural Network for Crowd Counting](https://mlsanity.com/paper/2208.02894/) (2022)
- [Revisiting Crowd Counting: State-of-the-art, Trends, and Future Perspectives](https://mlsanity.com/paper/2209.07271/) (2022)
- [Crowd Counting via Perspective-Guided Fractional-Dilation Convolution](https://mlsanity.com/paper/2107.03665/) (2021)
- [Variational Attention: Propagating Domain-Specific Knowledge for Multi-Domain Learning in Crowd Counting](https://mlsanity.com/paper/2108.08023/) (2021)
- [CountFormer: Multi-View Crowd Counting Transformer](https://mlsanity.com/paper/2407.02047/) (2024)
- [WheatNet: A Lightweight Convolutional Neural Network for High-throughput Image-based Wheat Head Detection and Counting](https://mlsanity.com/paper/2103.09408/) (2021)
- [Counting Varying Density Crowds Through Density Guided Adaptive Selection CNN and Transformer Estimation](https://mlsanity.com/paper/2206.10075/) (2022)
- [VisDrone-CC2020: The Vision Meets Drone Crowd Counting Challenge Results](https://mlsanity.com/paper/2107.08766/) (2021)
- [Weakly-Supervised Cross-Domain Segmentation of Electron Microscopy with Sparse Point Annotation](https://mlsanity.com/paper/2404.00667/) (2024)
- [Towards More Effective PRM-based Crowd Counting via A Multi-resolution Fusion and Attention Network](https://mlsanity.com/paper/2112.09664/) (2021)
- [Domain Randomization for Object Counting](https://mlsanity.com/paper/2202.08670/) (2022)
- [BBA-net: A bi-branch attention network for crowd counting](https://mlsanity.com/paper/2201.08983/) (2022)
- [Automatic counting of planting microsites via local visual detection and global count estimation](https://mlsanity.com/paper/2311.00796/) (2023)
- [Virtual Classification: Modulating Domain-Specific Knowledge for Multidomain Crowd Counting](https://mlsanity.com/paper/2402.03758/) (2024)
- [Bio-inspired visual relative localization for large swarms of UAVs](https://mlsanity.com/paper/2412.02393/) (2024)
- [Counting in the 2020s: Binned Representations and Inclusive Performance Measures for Deep Crowd Counting Approaches](https://mlsanity.com/paper/2204.04653/) (2022)
- [Tolerating Annotation Displacement in Dense Object Counting via Point Annotation Probability Map](https://mlsanity.com/paper/2308.00530/) (2023)
- [Spatially-Attentive Patch-Hierarchical Network with Adaptive Sampling for Motion Deblurring](https://mlsanity.com/paper/2402.06117/) (2024)
- [RPEE-HEADS: A Novel Benchmark for Pedestrian Head Detection in Crowd Videos](https://mlsanity.com/paper/2411.18164/) (2024)

## Abstract (arXiv)

The task of crowd counting is to automatically estimate the pedestrian number in crowd images. To cope with the scale and perspective changes that commonly exist in crowd images, state-of-the-art approaches employ multi-column CNN architectures to regress density maps of crowd images. Multiple columns have different receptive fields corresponding to pedestrians (heads) of different scales. We instead propose a scale-adaptive CNN (SaCNN) architecture with a backbone of fixed small receptive fields. We extract feature maps from multiple layers and adapt them to have the same output size; we combine them to produce the final density map. The number of people is computed by integrating the density map. We also introduce a relative count loss along with the density map loss to improve the network generalization on crowd scenes with few pedestrians, where most representative approaches perform poorly on. We conduct extensive experiments on the ShanghaiTech, UCF_CC_50 and WorldExpo datasets as well as a new dataset SmartCity that we collect for crowd scenes with few people. The results demonstrate significant improvements of SaCNN over the state-of-the-art.