Publications

Conference2025 IEEE International Conference on Multimedia and Expo (ICME)

CLIP-EBC: CLIP Can Count Accurately through Enhanced Blockwise Classification

Yiming Ma, Victor Sanchez, Tanaya Guha

Overview figure from the paper “CLIP-EBC: CLIP Can Count Accurately through Enhanced Blockwise Classification”

TL;DR

Classification-based crowd counters quantise counts into real-valued bins that clash with CLIP's pretraining corpus and blur labels near bin borders; EBC uses integer-valued bins plus a density-map loss, improving such methods by up to 44.5% in MAE on UCF-QNRF, and CLIP-EBC sets the state of the art on the NWPU-Crowd test set (MAE 58.2).

Abstract

We propose CLIP-EBC, the first fully CLIP-based model designed for accurate crowd density estimation. Existing classification-based crowd counting frameworks face challenges when directly applied with CLIP. For instance, these methods quantize count values into bordering real-valued bins, which are inconsistent with CLIP’s pretraining corpus. Besides, this quantization strategy also introduces label ambiguity near shared bin borders. To address these issues, we propose the Enhanced Blockwise Classification (EBC) framework, which utilizes integer-valued bins and additionally incorporates a density-map-based loss to further improve count accuracy. Building on this backbone-agnostic framework, CLIP-EBC fully leverages CLIP’s recognition capabilities for crowd counting. Experiments show that EBC improves classification-based methods by up to 44.5% in mean absolute error (MAE) on the UCF-QNRF dataset. Furthermore, CLIP-EBC achieves state-of-the-art performance on the NWPU-Crowd test set, with an MAE of 58.2, surpassing the previous best method by 8.6%. Our code is available at https://github.com/Yiming-M/CLIP-EBC.

Cite

@INPROCEEDINGS{11209839,
  author={Ma, Yiming and Sanchez, Victor and Guha, Tanaya},
  booktitle={2025 IEEE International Conference on Multimedia and Expo (ICME)},
  title={CLIP-EBC: CLIP Can Count Accurately through Enhanced Blockwise Classification},
  year={2025},
  pages={1-6},
  doi={10.1109/ICME59968.2025.11209839}}