Research map / Detection and segmentation / Open-set and few-shot detection
YOLOE: Real-Time Seeing Anything
Ao Wang, Lihao Liu, Hui Chen, Zijia Lin, Jungong Han, Guiguang Ding
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 RepRTA auxiliary text-embedding network (re-parameterized at inference) (ablation) | YOLOE-v8-L · LVIS minival zero-shot | AP: 31.5 -> 33.5 | Improved |
| remove cross-modality fusion for speed (ablation) | YOLO-Worldv2-L · LVIS minival zero-shot | AP: 31.9 -> 30.0 | Hurt |
| add global negative dictionary instead of empty-string negatives (ablation) | YOLO-Worldv2-L (30 epochs) · LVIS minival zero-shot | AP: 31.0 -> 31.9 | Improved |
| use MobileCLIP-B(LT) text encoder (ablation) | YOLOE without cross-modal fusion · LVIS minival zero-shot | AP: 30.0 -> 31.5 | Improved |
| reduce training epochs from 100 to 30 (ablation) | YOLO-Worldv2-L · LVIS minival zero-shot | AP: 33.0 -> 31.0 | Hurt |
| add segmentation head and train detection and segmentation jointly (ablation) | YOLOE-v8-L · LVIS minival zero-shot | AP: 33.5 -> 33.3 | Hurt |
| replace mask pooling with SAVPE activation branch for visual prompts (ablation) | YOLOE visual prompt · LVIS minival zero-shot (visual prompt) | AP: +1.5 AP | Improved |
| increase SAVPE group number A from 1 to 16 (ablation) | SAVPE · LVIS minival zero-shot (visual prompt) | AP: 31.9 AP at A=16 | No clear effect |
| use LRPC lazy retrieval instead of matching full vocabulary against all anchor points (ablation) | YOLOE prompt-free · LVIS minival prompt-free | inference speedup: 1.7x / 1.3x speedup for YOLOE-v8-S / L, same AP | Improved |
| raise LRPC filtering threshold delta for more speed (ablation) | YOLOE-v8-S LRPC · LVIS minival prompt-free | AP: 1.9x speedup with 0.2 AP drop | Mixed |
| fine-tune YOLOE fully on COCO with far fewer epochs instead of training YOLOv8-L from scratch | YOLOE-v8-L · COCO val | AP^b: 52.4 -> 53.0 | Improved |
| joint detection and segmentation training in YOLOE vs YOLO-Worldv2 on frequent categories | YOLOE-v8-M/L · LVIS minival zero-shot | AP_f: YOLOE-v8-L 37.3 vs YOLO-Worldv2-L 38.1 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- T-Rex2 — T-Rex2: Towards Generic Object Detection via Text-Visual Prompt Synergy (ECCV 2024) · better on 0 of 1 shared result
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
- YOLO-World: Real-Time Open-Vocabulary Object Detection (CVPR 2024 · similarity 0.89)
- Just a Few Glances: Open-Set Visual Perception with Image Prompt Paradigm (AAAI 2025 · similarity 0.89)
- YOLO-RD: Introducing Relevant and Compact Explicit Knowledge to YOLO by Retriever-Dictionary (ICLR 2025 · similarity 0.88)
- Roboflow100-VL: A Multi-Domain Object Detection Benchmark for Vision-Language Models (NeurIPS 2025 · similarity 0.88)
- Visual Diversity and Region-aware Prompt Learning for Zero-shot HOI Detection (NeurIPS 2025 · similarity 0.87)
- DitHub: A Modular Framework for Incremental Open-Vocabulary Object Detection (NeurIPS 2025 · similarity 0.87)
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.
