mlsanityData to 14 Oct 2026

Spatially-Aware Class-Agnostic Object Counting

arXiv 2607.16826 ↗First posted 18 Jul 2026cs.CVcited by 1

Not on the map. Why?

Abstract (arXiv)

Generalised object counting aims to estimate the number of instances of an arbitrary object category from a single image, but many recent methods can struggle on structurally complex objects due to limited spatial modelling. We present textitUpCount, a class-agnostic counter designed to better preserve spatial structure. UpCount strengthens the visual representation by extracting multi-layer features from a ViT-B/16 encoder and reassembling them into a refined multi-scale pyramid that is spatially refined using Dense Prediction Transformers and FeatUp, yielding features with improved structural and spatial sensitivity; a proposal--verification counting head then identifies repeated patterns and produces a density map for the final count. On FSC-147, UpCount achieves 12.39 MAE and 100.89 RMSE on the test set, and it transfers effectively to vehicle counting on CARPK (6.27 MAE, 8.79 RMSE). Code: https://github.com/r28112072-rgb/upcount

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 18 of its 20 references that are in the map. First posted on arXiv 18 Jul 2026.

Builds on

18 of 20 references.

References · 18

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2021Vision Transformers for Dense Predictioncited by 1,803
  3. 2017Drone-based Object Counting by Spatially Regularized Regional Proposal Networkcited by 156
  4. 2021Learning To Count Everythingcited by 150
  5. 2024FeatUp: A Model-Agnostic Framework for Features at Any Resolutioncited by 116
  6. 2016A Large Contextual Dataset for Classification, Detection and Counting of Cars with Deep Learningcited by 86
  7. 2018Class-Agnostic Countingcited by 85
  8. 2023CLIP-Count: Towards Text-Guided Zero-Shot Object Countingcited by 83
  9. 2022CounTR: Transformer-based Generalised Visual Countingcited by 79
  10. 2022A Low-Shot Object Counting Network With Iterative Prototype Adaptationcited by 65
  11. 2024CountGD: Multi-Modal Open-World Countingcited by 59
  12. 2023Zero-shot Object Countingcited by 57
6 more
  1. 2024DAVE -- A Detect-and-Verify Paradigm for Low-Shot Countingcited by 49
  2. 2023Open-world Text-specified Object Countingcited by 41
  3. 2022Exemplar Free Class Agnostic Countingcited by 32
  4. 2025T2ICount: Enhancing Cross-modal Understanding for Zero-Shot Countingcited by 19
  5. 2025CountingDINO: A Training-free Pipeline for Class-Agnostic Counting using Unsupervised Backbonescited by 6
  6. 2026Boosting Quantitive and Spatial Awareness for Zero-Shot Object Countingcited by 5

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Object Counting Across Modalities: Taxonomies, Benchmarks, Applications, and Open Challengescited by 0