Research map / 3D vision / 3D generation
BoxCars: 3D Boxes as CNN Input for Improved Fine-Grained Vehicle Recognition
Check 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 |
|---|---|---|---|
| Use unpacked vehicle images with rasterized box and viewpoint inputs | BVLC Reference CaffeNet on BoxCars · BoxCars classification, medium, three samples per vehicle | Top-1 accuracy: 0.772 -> 0.832 | Helped |
| Use unpacked vehicle images with rasterized box and viewpoint inputs | BVLC Reference CaffeNet on BoxCars · BoxCars classification, hard, three samples per vehicle | Top-1 accuracy: 0.733 -> 0.804 | Helped |
| Use unpacked vehicle images with rasterized box and viewpoint inputs | BVLC Reference CaffeNet on BoxCars · BoxCars verification, medium, three samples per vehicle | average precision: 0.378 -> 0.785 | Helped |
| Use unpacked vehicle images with rasterized box and viewpoint inputs | BVLC Reference CaffeNet on BoxCars · BoxCars verification, hard, three samples per vehicle | average precision: 0.353 -> 0.710 | Helped |
| Add unpacking of the vehicle image (ablation) | BVLC Reference CaffeNet on BoxCars · BoxCars classification, medium, one sample per vehicle | Top-1 accuracy: 0.720 -> 0.764 | Helped |
| Add unpacking of the vehicle image (ablation) | BVLC Reference CaffeNet on BoxCars · BoxCars classification, hard, one sample per vehicle | Top-1 accuracy: 0.669 -> 0.730 | Helped |
| Add rasterized bounding-box input (ablation) | BVLC Reference CaffeNet on BoxCars · BoxCars classification, medium, one sample per vehicle | Top-1 accuracy: 0.720 -> 0.748 | Helped |
| Add rasterized bounding-box input (ablation) | BVLC Reference CaffeNet on BoxCars · BoxCars classification, hard, one sample per vehicle | Top-1 accuracy: 0.669 -> 0.694 | Helped |
| Add encoded viewpoint input (ablation) | BVLC Reference CaffeNet on BoxCars · BoxCars classification, medium, one sample per vehicle | Top-1 accuracy: 0.720 -> 0.712 | No clear effect |
| Add encoded viewpoint input (ablation) | BVLC Reference CaffeNet on BoxCars · BoxCars classification, hard, one sample per vehicle | Top-1 accuracy: 0.669 -> 0.676 | No clear effect |
| Add unpacking, rasterized box and viewpoint inputs (ablation) | BVLC Reference CaffeNet on BoxCars · BoxCars classification, medium, one sample per vehicle | Top-1 accuracy: 0.720 -> 0.754 | Helped |
| Add unpacked image and rasterized bounding-box input (ablation) | BVLC Reference CaffeNet on BoxCars · BoxCars verification, medium, three samples per vehicle | average precision: 0.378 -> 0.771 | Helped |
Closest papers by meaning
- Geometry-Constrained Car Recognition Using a 3D Perspective Network (AAAI 2020 · similarity 0.87)
- 3D Part Guided Image Editing for Fine-Grained Object Understanding (CVPR 2020 · similarity 0.86)
- Stereo R-CNN Based 3D Object Detection for Autonomous Driving (CVPR 2019 · similarity 0.86)
- OpenBox: Annotate Any Bounding Boxes in 3D (NeurIPS 2025 · similarity 0.84)
- DI-V2X: Learning Domain-Invariant Representation for Vehicle-Infrastructure Collaborative 3D Object Detection (AAAI 2024 · similarity 0.83)
- Structural Information Guided Multimodal Pre-training for Vehicle-Centric Perception (AAAI 2024 · similarity 0.82)
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.