Research map / Detection and segmentation / Open-set and few-shot detection
Multi-modal Queried Object Detection in the Wild
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| set mask rate to 0 (no masked language prediction) (ablation) | MQ-GLIP-T · LVIS MiniVal | AP: 30.4 -> 26.9 | Hurt |
| raise mask rate to 80% (ablation) | MQ-GLIP-T · LVIS MiniVal | AP: 30.4 -> 28.9 | Hurt |
| remove the gate in GCP (ablation) | MQ-GLIP-T · LVIS MiniVal | AP: 30.4 -> 27.1 | Hurt |
| replace MLP gate with a single learnable scaler (ablation) | MQ-GLIP-T · LVIS MiniVal | AP: 30.4 -> 29.3 | Hurt |
| replace MLP gate with one linear layer (ablation) | MQ-GLIP-T · LVIS MiniVal | AP: 30.4 -> 30.0 | Hurt |
| feed cat(v,t) to the gate instead of v (ablation) | MQ-GLIP-T · LVIS MiniVal | AP: 30.4 -> 29.2 | Hurt |
| unfreeze the whole pre-trained detector (ablation) | MQ-GLIP-T · LVIS MiniVal | AP: 30.4 -> 24.4 | Hurt |
| unfreeze the text encoder (ablation) | MQ-GLIP-T · LVIS MiniVal | AP: 30.4 -> 26.6 | Hurt |
| replace text queries with [MASK], vision queries only (ablation) | MQ-GLIP-T · LVIS MiniVal | AP: 30.4 -> 17.6 | Hurt |
| add vision queries to GLIP-T (finetuning-free) | GLIP-T · LVIS MiniVal | AP: 26.0 -> 30.4 | Helped |
| add vision queries to GLIP-T (finetuning-free) | GLIP-T · ODinW-13 | AP_avg: 41.9 -> 45.6 | Helped |
| add vision queries to GLIP-T (3-shot) | GLIP-T · ODinW-13 | AP_avg: 50.7 -> 57.0 | Helped |
Closest papers by meaning
- Cyclic Contrastive Knowledge Transfer for Open-Vocabulary Object Detection (ICLR 2025 · similarity 0.89)
- Comprehensive Multi-Modal Prototypes Are Simple and Effective Classifiers for Vast-Vocabulary Object Detection (AAAI 2025 · similarity 0.88)
- Analyzing and Boosting the Power of Fine-Grained Visual Recognition for Multi-modal Large Language Models (ICLR 2025 · similarity 0.88)
- Roboflow100-VL: A Multi-Domain Object Detection Benchmark for Vision-Language Models (NeurIPS 2025 · similarity 0.88)
- ROD-MLLM: Towards More Reliable Object Detection in Multimodal Large Language Models (CVPR 2025 · similarity 0.87)
- Exploring Region-Word Alignment in Built-in Detector for Open-Vocabulary Object Detection (CVPR 2024 · 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.