# ZIP: Scalable Crowd Counting via Zero-Inflated Poisson Modeling

arXiv: https://arxiv.org/abs/2506.19955 · HTML: https://mlsanity.com/paper/2506.19955/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 19 of its 31 references that are in the map.
- First posted on arXiv 24 Jun 2025.

## Builds on (most cited first, 19 of 19)

- [Reproducible scaling laws for contrastive language-image learning](https://mlsanity.com/paper/2212.07143/) (2022)
- [Composition Loss for Counting, Density Map Estimation and Localization in Dense Crowds](https://mlsanity.com/paper/1808.01050/) (2018)
- [Bayesian Loss for Crowd Count Estimation with Point Supervision](https://mlsanity.com/paper/1908.03684/) (2019)
- [NWPU-Crowd: A Large-Scale Benchmark for Crowd Counting and Localization](https://mlsanity.com/paper/2001.03360/) (2020)
- [Context-Aware Crowd Counting](https://mlsanity.com/paper/1811.10452/) (2018)
- [Distribution Matching for Crowd Counting](https://mlsanity.com/paper/2009.13077/) (2020)
- [MobileNetV4 -- Universal Models for the Mobile Ecosystem](https://mlsanity.com/paper/2404.10518/) (2024)
- [MobileCLIP: Fast Image-Text Models through Multi-Modal Reinforced Training](https://mlsanity.com/paper/2311.17049/) (2023)
- [Boosting Crowd Counting via Multifaceted Attention](https://mlsanity.com/paper/2203.02636/) (2022)
- [Rethinking Spatial Invariance of Convolutional Networks for Object Counting](https://mlsanity.com/paper/2206.05253/) (2022)
- [STEERER: Resolving Scale Variations for Counting and Localization via Selective Inheritance Learning](https://mlsanity.com/paper/2308.10468/) (2023)
- [Point-Query Quadtree for Crowd Counting, Localization, and More](https://mlsanity.com/paper/2308.13814/) (2023)
- [Uniformity in Heterogeneity:Diving Deep into Count Interval Partition for Crowd Counting](https://mlsanity.com/paper/2107.12619/) (2021)
- [Gramformer: Learning Crowd Counting via Graph-Modulated Transformer](https://mlsanity.com/paper/2401.03870/) (2024)
- [Regressor-Segmenter Mutual Prompt Learning for Crowd Counting](https://mlsanity.com/paper/2312.01711/) (2023)
- [Boosting Detection in Crowd Analysis via Underutilized Output Features](https://mlsanity.com/paper/2308.16187/) (2023)
- [CLIP-EBC: CLIP Can Count Accurately through Enhanced Blockwise Classification](https://mlsanity.com/paper/2403.09281/) (2024)
- [Coarse- and Fine-grained Attention Network with Background-aware Loss for Crowd Density Map Estimation](https://mlsanity.com/paper/2011.03721/) (2020)
- [Understanding the impact of mistakes on background regions in crowd counting](https://mlsanity.com/paper/2003.13759/) (2020)

## Built on by (most cited first, 2 of 2)

- [Dense Point-to-Mask Optimization with Reinforced Point Selection for Crowd Instance Segmentation](https://mlsanity.com/paper/2604.01742/) (2026)
- [CrowdCue: Specialist-Cue Conditioning for Vision-Language Crowd Counting](https://mlsanity.com/paper/2609.23012/) (2026)

## Abstract (arXiv)

Most crowd counting methods directly regress blockwise density maps using Mean Squared Error (MSE) losses. This practice has two key limitations: (1) it fails to account for the extreme spatial sparsity of annotations - over 95% of 8x8 blocks are empty across standard benchmarks, so supervision signals in informative regions are diluted by the predominant zeros; (2) MSE corresponds to a Gaussian error model that poorly matches discrete, non-negative count data. To address these issues, we introduce ZIP, a scalable crowd counting framework that models blockwise counts with a Zero-Inflated Poisson likelihood: a zero-inflation term learns the probability a block is structurally empty (handling excess zeros), while the Poisson component captures expected counts when people are present (respecting discreteness). We provide a generalization analysis showing a tighter risk bound for ZIP than MSE-based losses and DMCount provided that the training resolution is moderately large. To assess the scalability of ZIP, we instantiate it on backbones spanning over 100x in parameters/compute. Experiments on ShanghaiTech A & B, UCF-QNRF, and NWPU-Crowd demonstrate that ZIP consistently surpasses state-of-the-art methods across all model scales.