Research map / Detection and segmentation / Open-set and few-shot detection
Object-level Semantic and Spatial Distillation for Open Vocabulary Detection
Zitong Li, Jinzhuo Wu, Fukang Zhao, Xinyue Wang, Jun Chen-CUG, Zhuo Cheng, Dapeng Luo
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 spatial distillation (ablation) | DINO-DETR ViT-L/14 · OV-COCO | AP50 novel: 38.3 -> 42.3 | Improved |
| add semantic distillation (ablation) | DINO-DETR ViT-L/14 · OV-COCO | AP50 novel: 38.3 -> 45.0 | Improved |
| add LQEH (ablation) | DINO-DETR ViT-L/14 · OV-COCO | AP50 novel: 38.3 -> 41.7 | Improved |
| add spatial distillation to semantic distillation (ablation) | DINO-DETR ViT-L/14 · OV-COCO | AP50 novel: 45.0 -> 47.8 | Improved |
| add LQEH to both distillations (ablation) | DINO-DETR ViT-L/14 · OV-COCO | AP50 novel: 47.8 -> 49.2 | Improved |
| swap DETR encoder teacher for DINOv2 as spatial teacher (ablation) | OSSD ViT-L/14 · OV-COCO | AP50 novel: 49.2 -> 52.4 | Improved |
| swap DETR encoder teacher for DINOv2 as spatial teacher (ablation) | OSSD ViT-L/14 · OV-LVIS | mAPr: 40.5 -> 41.8 | Improved |
| replace last attention with v-v attention (ablation) | CLIP ViT-L/14 · OV-COCO region classification | Top-1 accuracy (all): 40.9 -> 69.4 | Improved |
| use v-v instead of k-k last attention (ablation) | CLIP ViT-L/14 · OV-COCO region classification | Top-1 accuracy (all): 66.1 -> 69.4 | Improved |
| add token prompt on last 5 layers instead of final (ablation) | CLIP ViT-L/14 v-v · OV-COCO region classification | Top-1 novel accuracy: 72.2 -> 73.6 | Improved |
| add semantic distillation on all layers vs last 3 (ablation) | CLIP ViT-L/14 v-v · OV-COCO region classification | Top-1 accuracy (all): 71.0 -> 77.2 | Improved |
| add semantic distillation (all layers) versus token prompt last 5 (ablation) | CLIP ViT-L/14 v-v · OV-COCO region classification | Top-1 base accuracy: 88.2 -> 76.3 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 4 of 4 on most shared results.
- OV-DQUO — OV-DQUO: Open-Vocabulary DETR with Denoising Text Query Training and Open-World Unknown Objects Supervision (AAAI 2025) · better on 3 of 3 shared results
- ProxyDet — ProxyDet: Synthesizing Proxy Novel Classes via Classwise Mixup for Open-Vocabulary Object Detection (AAAI 2024) · better on 1 of 1 shared result
- CORA+ — CORA: Adapting CLIP for Open-Vocabulary Detection With Region Prompting and Anchor Pre-Matching (CVPR 2023) · better on 1 of 1 shared result
- BIND — BIND: Binary Integrated Net Descriptors for Texture-Less Object Recognition (CVPR 2017) · better on 1 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
- Explicit Semantic–Spatial Alignment for Open-Vocabulary Object Detection (ECCV 2026 · similarity 0.93)
- CAKE: Category Aware Knowledge Extraction for Open-Vocabulary Object Detection (AAAI 2025 · similarity 0.90)
- Exploring Region-Word Alignment in Built-in Detector for Open-Vocabulary Object Detection (CVPR 2024 · similarity 0.90)
- Cyclic Contrastive Knowledge Transfer for Open-Vocabulary Object Detection (ICLR 2025 · similarity 0.89)
- LLMs Meet VLMs: Boost Open Vocabulary Object Detection with Fine-grained Descriptors (ICLR 2024 · similarity 0.89)
- Distilling Spectral Graph for Object-Context Aware Open-Vocabulary Semantic Segmentation (CVPR 2025 · 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.
