Research map / Vision-language and multimodal / Cross-modal learning
LapsCore: Language-Guided Person Search via Color Reasoning
Yushuang Wu, Zizheng Yan, Xiaoguang Han, Guanbin Li, Changqing Zou, Shuguang Cui
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add TC and IC color reasoning tasks | CMPM/C · CUHK-PEDES | Recall@1: 49.37 -> 53.33 | Improved |
| add TC and IC color reasoning tasks | CMP adv · CUHK-PEDES | Recall@1: 55.05 -> 57.00 | Improved |
| add TC and IC color reasoning tasks | NAFS · CUHK-PEDES | Recall@1: 61.50 -> 63.40 | Improved |
| add TC and IC color reasoning tasks | NAFS · CUHK-PEDES | Recall@10: 87.51 -> 87.80 | Improved |
| add text completion (TC) only (ablation) | CMPM/C · CUHK-PEDES | Recall@1: 49.4 -> 51.8 | Improved |
| add image colorization (IC) only (ablation) | CMPM/C · CUHK-PEDES | Recall@1: 49.4 -> 52.5 | Improved |
| add IC on top of TC (TC&IC versus TC only) (ablation) | CMPM/C · CUHK-PEDES | Recall@1: 51.8 -> 53.3 | Improved |
| use grayscale instead of sketch images as colorization source (ablation) | IC · CUHK-PEDES | Recall@1: 51.5 -> 52.5 | Improved |
| fill color words instead of object words in text completion (ablation) | TC · CUHK-PEDES | Recall@1: 50.1 -> 51.8 | Improved |
| complete shallow conv1 features instead of deeper conv2_x features (ablation) | ICf · CUHK-PEDES | Recall@1: 51.2 -> 52.7 | Improved |
| mask one in two channels instead of one in four (ablation) | ICf · CUHK-PEDES | Recall@1: 51.7 -> 52.7 | Improved |
| mask channels instead of spatial regions (ablation) | ICf · CUHK-PEDES | Recall@1: 51.6 -> 52.7 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ViTAA — ViTAA: Visual-Textual Attributes Alignment in Person Search by Natural Language (ECCV 2020) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- PLIP: Language-Image Pre-training for Person Representation Learning (NeurIPS 2024 · similarity 0.87)
- An Empirical Study of CLIP for Text-Based Person Search (AAAI 2024 · similarity 0.86)
- ColorConceptBench: A Benchmark for Probabilistic Color-Concept Understanding in Text-to-Image Models (NeurIPS 2026 · similarity 0.86)
- Human-centered Interactive Learning via MLLMs for Text-to-Image Person Re-identification (CVPR 2025 · similarity 0.85)
- Beyond Global Similarity: Multi-Conditional Retrieval for Fine-Grained Cross-Modal Understanding (CVPR 2026 · similarity 0.84)
- Visual Perturbation for Text-Based Person Search (AAAI 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
