Research map / Detection and segmentation / Open-set and few-shot detection
LaMI-DETR: Open-Vocabulary Detection with Language Model Instruction
Penghui Du, Yu Wang, Yifan Sun, Luting Wang, Yue Liao, gang zhang, Errui Ding, Yan Wang, Jingdong Wang, Si Liu
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 language embedding fusion to the detector queries (ablation) | LaMI-DETR · OV-LVIS | AP_r: 32.2 -> 33.0 | Improved |
| add visual concept sampling (cluster visual descriptions, sample easy negatives) (ablation) | LaMI-DETR · OV-LVIS | AP_r: 33.0 -> 40.1 | Improved |
| add embedding update with visual descriptions (ablation) | LaMI-DETR · OV-LVIS | AP_r: 40.1 -> 42.5 | Improved |
| add confusing-category prompts at inference (ablation) | LaMI-DETR · OV-LVIS | AP_r: 42.5 -> 43.4 | Improved |
| regenerate descriptions to emphasize differences from the most confusable category (ablation) | CLIP text classifier · LVIS minival proposal classification | mAcc_r: 49.5 -> 52.7 | Improved |
| cluster with CLIP text name embeddings for visual concept sampling (ablation) | baseline+VCS · OV-LVIS | AP_r: 33.0 -> 33.5 | No clear effect |
| cluster T5 name embeddings instead of CLIP (ablation) | baseline+VCS · OV-LVIS | AP_r: 33.5 -> 34.1 | No clear effect |
| cluster name plus definition embeddings (DetCLIP-style enrichment) (ablation) | baseline+VCS · OV-LVIS | AP_r: 33.0 -> 31.5 | Hurt |
| cluster name plus visual description embeddings (ablation) | baseline+VCS · OV-LVIS | AP_r: 34.1 -> 40.1 | Improved |
| multiply novel-class logits by factor 5.0 at inference (ablation) | baseline+LaMI · OV-LVIS | AP_r: 41.7 -> 43.4 AP_r; AP 41.1 -> 41.3 | Mixed |
| LaMI-DETR vs strongest prior OV-LVIS method | LaMI-DETR · OV-LVIS | AP_r: 35.6 -> 43.4 | Improved |
Later papers that compared with this method
1 later accepted paper reports LaMI-DETR as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- GUIDED: Granular Understanding via Identification, Detection, and Discrimination for Fine-Grained Open-Vocabulary Object Detection (NeurIPS 2025 · better on 3 of 3 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
- DetCLIPv3: Towards Versatile Generative Open-vocabulary Object Detection (CVPR 2024 · similarity 0.90)
- Cyclic Contrastive Knowledge Transfer for Open-Vocabulary Object Detection (ICLR 2025 · similarity 0.90)
- LLMs Meet VLMs: Boost Open Vocabulary Object Detection with Fine-grained Descriptors (ICLR 2024 · similarity 0.89)
- Open-Det: An Efficient Learning Framework for Open-Ended Detection (ICML 2025 · similarity 0.89)
- Exploring Region-Word Alignment in Built-in Detector for Open-Vocabulary Object Detection (CVPR 2024 · similarity 0.89)
- OVMR: Open-Vocabulary Recognition with Multi-Modal References (CVPR 2024 · similarity 0.89)
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.
