Research map / 3D vision / Human pose estimation
Latent-Class Hough Forests for 6 DoF Object Pose Estimation
Alykhan Tejani, Rigas Kouskouridas, Andreas Doumanoglou, Danhang Tang, Tae‐Kyun Kim
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 iterative latent-class co-training at inference (ablation) | Latent-Class Hough Forests · Hinterstoisser LINEMOD dataset (average) | F1-score: 0.788 -> 0.817 | Improved |
| add iterative latent-class co-training at inference (ablation) | Latent-Class Hough Forests · Domestic Environment Dataset (average) | F1-score: 0.582 -> 0.633 | Improved |
| add iterative co-training (ablation) | Latent-Class Hough Forests · Hinterstoisser LINEMOD dataset Box | F1-score: 0.736 -> 0.740 | No clear effect |
| use scale-invariant LINEMOD templates instead of holistic LINEMOD | LINEMOD · Hinterstoisser LINEMOD dataset (average) | F1-score: 0.630 -> 0.702 | Improved |
| use latent-class forest instead of scale-invariant LINEMOD templates | template matching · Hinterstoisser LINEMOD dataset (average) | F1-score: 0.702 -> 0.788 | Improved |
| add depth to RGB input (ablation) | Latent-Class Hough Forests · Validation subset Coffee Cup | F1-score: 0.838 -> 0.828 | No clear effect |
| add depth to RGB input (ablation) | Latent-Class Hough Forests · Validation subset Camera | F1-score: 0.309 -> 0.369 | Improved |
| use larger forest and patch (10 trees, 2/3 patch) instead of 5 trees, 1/2 patch (ablation) | Latent-Class Hough Forests · Domestic Environment Dataset (average, top hypothesis) | accurate detections (%): 77.6% -> 80.0% | Improved |
| use larger forest and patch (10 trees, 2/3 patch) instead of 5 trees, 1/2 patch (ablation) | Latent-Class Hough Forests · Bin-Picking Dataset (average, top hypothesis) | accurate detections (%): 89.8% -> 90.8% | Improved |
| increase patch size toward holistic (4/5 and 1) (ablation) | Latent-Class Hough Forests · validation set | F1-score | Hurt |
| add negative training data (ablation) | Latent-Class Hough Forests · validation set | accuracy | No clear effect |
| use more than ten trees (ablation) | Latent-Class Hough Forests with patch >= 2/3 · validation set | F1-score | No clear effect |
Closest papers by meaning
- Recovering 6D Object Pose and Predicting Next-Best-View in the Crowd (CVPR 2016 · similarity 0.88)
- HoughNet: Integrating near and long-range evidence for bottom-up object detection (ECCV 2020 · similarity 0.87)
- HoughNet: Integrating Near and Long-Range Evidence for Visual Detection (TPAMI 2022 · similarity 0.87)
- Omni6D: Large-Vocabulary 3D Object Dataset for Category-Level 6D Object Pose Estimation (ECCV 2024 · similarity 0.85)
- BB8: A Scalable, Accurate, Robust to Partial Occlusion Method for Predicting the 3D Poses of Challenging Objects Without Using Depth (ICCV 2017 · similarity 0.85)
- ImOV3D: Learning Open Vocabulary Point Clouds 3D Object Detection from Only 2D Images (NeurIPS 2024 · similarity 0.84)
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.
