# DCA-MoE: Spatially Adaptive Cross-Layer Fusion and Density-Routed Experts for Crowd Counting

arXiv: https://arxiv.org/abs/2608.15213 · HTML: https://mlsanity.com/paper/2608.15213/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Vision mixture-of-experts, one of 12 directions in Mixture-of-experts.
- Builds on 27 of its 33 references that are in the map.
- First posted on arXiv 15 Aug 2026.

## Placed in

- Mixture-of-experts / [Vision mixture-of-experts](https://mlsanity.com/moe/vision-mixture-of-experts/)

## Builds on (most cited first, 27 of 27)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [Emerging Properties in Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.14294/) (2021)
- [Feature Pyramid Networks for Object Detection](https://mlsanity.com/paper/1612.03144/) (2016)
- [A ConvNet for the 2020s](https://mlsanity.com/paper/2201.03545/) (2022)
- [Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer](https://mlsanity.com/paper/1701.06538/) (2017)
- [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://mlsanity.com/paper/2101.03961/) (2021)
- [DINOv3](https://mlsanity.com/paper/2508.10104/) (2025)
- [Scaling Vision with Sparse Mixture of Experts](https://mlsanity.com/paper/2106.05974/) (2021)
- [Dynamic Convolution: Attention over Convolution Kernels](https://mlsanity.com/paper/1912.03458/) (2019)
- [CSRNet: Dilated Convolutional Neural Networks for Understanding the Highly Congested Scenes](https://mlsanity.com/paper/1802.10062/) (2018)
- [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)
- [Rethinking Counting and Localization in Crowds:A Purely Point-Based Framework](https://mlsanity.com/paper/2107.12746/) (2021)
- [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)
- [CLIP-Count: Towards Text-Guided Zero-Shot Object Counting](https://mlsanity.com/paper/2305.07304/) (2023)
- [An End-to-End Transformer Model for Crowd Localization](https://mlsanity.com/paper/2202.13065/) (2022)
- [Boosting Crowd Counting via Multifaceted Attention](https://mlsanity.com/paper/2203.02636/) (2022)
- [From Open Set to Closed Set: Counting Objects by Spatial Divide-and-Conquer](https://mlsanity.com/paper/1908.06473/) (2019)
- [CrowdCLIP: Unsupervised Crowd Counting via Vision-Language Model](https://mlsanity.com/paper/2304.04231/) (2023)
- [STEERER: Resolving Scale Variations for Counting and Localization via Selective Inheritance Learning](https://mlsanity.com/paper/2308.10468/) (2023)
- [Regressor-Segmenter Mutual Prompt Learning for Crowd Counting](https://mlsanity.com/paper/2312.01711/) (2023)
- [FusionCount: Efficient Crowd Counting via Multiscale Feature Fusion](https://mlsanity.com/paper/2202.13660/) (2022)
- [CLIP-EBC: CLIP Can Count Accurately through Enhanced Blockwise Classification](https://mlsanity.com/paper/2403.09281/) (2024)

## Abstract (arXiv)

Crowd counting must recover reliable local density under severe variations in perspective, head scale, occlusion, and background clutter. Although modern counting objectives provide strong spatial supervision, many multi-level decoders still use spatially invariant feature fusion and apply one receptive-field pattern to every location. We propose DCA-MoE, a framework that makes both decisions content dependent while retaining a frozen DINOv3 encoder. Spatially Adaptive Layer Fusion (SALF) predicts position-wise weights over four aligned backbone features, and Density-Routed Multi-Receptive-Field Experts (DR-MoE) assigns each location a soft mixture of local, mid-range, and large-context residual experts. An EBC-style head reconstructs block density, while DMCount supervision and an auxiliary routing-balance term train the decoder without updating the backbone. On the NWPU-Crowd validation split, the strongest paired configuration, based on DINOv3 ViT-L/16, obtains 31.7 MAE and 72.2 RMSE; the matched ViT-B/16 full model obtains a paired 32.2/75.9. Cross-dataset results remain mixed, and several component baselines currently report independently selected minima from a single seed. The evidence therefore supports the feasibility of spatially adaptive fusion and routing, while broader paired and multi-seed evaluation remains necessary for causal attribution.