Research map / Detection and segmentation / Open-set and few-shot detection
DeepTag: A General Framework for Fiducial Marker Design and Detection
Zhuming Zhang, Yongtao Hu, Guoxing Yu, Jingwen Dai
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 |
|---|---|---|---|
| replace hand-crafted detection with DeepTag CNN keypoint detection | AprilTag (36h11) · New dataset Normal (<=50cm, <=60deg) | position error (mm): 0.390 -> 0.370 | Improved |
| replace hand-crafted detection with DeepTag | ArUco · New dataset Normal (<=50cm, <=60deg) | position error (mm): 3.088 -> 0.231 | Improved |
| replace hand-crafted detection with DeepTag | TopoTag · New dataset Difficult (>50cm or >60deg) | detection recall (%): 5.18 -> 59.86 | Improved |
| replace hand-crafted detection with DeepTag | RuneTag · Yu's dataset | detection recall (%): 0 -> 100 | Improved |
| replace hand-crafted detection with DeepTag | AprilTag · New dataset Difficult (>50cm or >60deg) | detection recall (%): 64.83 -> 76.14 | Improved |
| add ROI refinement with detected corners in Stage-1 (ablation) | DeepTag Stage-1 · New dataset, 30cm fixed distance | marker pose error | Improved |
| add Stage-2 ROI update on top of Stage-1 refined ROI (ablation) | DeepTag · New dataset, 30cm fixed distance | marker pose error | Improved |
| use all N x N keypoints instead of 4 corners or boundary points for pose (ablation) | DeepTag Stage-2 · New dataset | pose accuracy | Improved |
| add a second Stage-2 iteration (ablation) | DeepTag Stage-2 · New dataset | pose accuracy | Improved |
| use complex local patterns (AprilTag-XO) instead of plain AprilTag | DeepTag · New dataset Normal (<=50cm, <=60deg) | position error (mm): 0.370 -> 0.161 | Improved |
| use complex local patterns (AprilTag-XO) instead of plain AprilTag | DeepTag · New dataset Difficult (>50cm or >60deg) | detection recall (%): 76.14 -> 65.94 | Hurt |
| complex patterns at 10cm vs 30cm view distance | DeepTag AprilTag-XO/XA · New dataset 10cm and 30cm | pose accuracy | Mixed |
Closest papers by meaning
- An Accurate and Robust Artificial Marker Based on Cyclic Codes (TPAMI 2016 · similarity 0.87)
- ChromaTag: A Colored Marker and Fast Detection Algorithm (ICCV 2017 · similarity 0.86)
- Deep Virtual Markers for Articulated 3D Shapes (ICCV 2021 · similarity 0.85)
- Learnable Visual Markers (NeurIPS 2016 · similarity 0.84)
- SOMA: Solving Optical Marker-Based MoCap Automatically (ICCV 2021 · similarity 0.84)
- Tag2Text: Guiding Vision-Language Model via Image Tagging (ICLR 2024 · similarity 0.81)
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.
