Research map / 3D vision / Depth and stereo
FisheyeHDK: Hyperbolic Deformable Kernel Learning for Ultra-Wide Field-of-View Image Recognition
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 |
|---|---|---|---|
| Use hyperbolic rather than Euclidean deformable kernel positions | DeepLabV3 ResNet101 · Cityscapes fisheye f=200 | mIoU: 57.7 -> 59.6 | Helped |
| Use hyperbolic rather than Euclidean deformable kernel positions | DeepLabV3 ResNet101 · BDD100K fisheye f=75 | mIoU: 42.8 -> 44.0 | Helped |
| Use hyperbolic rather than Euclidean deformable kernel positions | DeepLabV3 ResNet101 · BDD100K fisheye f=75 | mAcc: 57.6 -> 56.6 | Hurt |
| Use hyperbolic rather than Euclidean deformable kernel positions | DeepLabV3 ResNet101 · Real fisheye test | mIoU: 76.3 -> 80.1 | Helped |
| Apply deformable kernels to first 3 ResNet layers rather than all layers (ablation) | FisheyeHDK · Cityscapes fisheye f=200 | mIoU: 58.0 -> 58.4 | Helped |
| Apply deformable kernels to last 3 ResNet layers rather than all layers (ablation) | FisheyeHDK · Cityscapes fisheye f=200 | mIoU: 58.0 -> 59.6 | Helped |
| Extend deformable kernels from the first 3 layers to the first 6 (ablation) | FisheyeHDK · Cityscapes fisheye f=200 | mIoU: 58.4 -> 58.5 | No clear effect |
| Extend deformable kernels from the last 3 layers to the last 6 (ablation) | FisheyeHDK · Cityscapes fisheye f=200 | mIoU: 59.6 -> 59.3 | Hurt |
| Increase graph input downsampling from factor 4 to factor 8 (ablation) | FisheyeHDK · Cityscapes fisheye efficiency comparison | test time (one image): 0.318s -> 0.288s | Helped |
Closest papers by meaning
- Fully Hyperbolic Convolutional Neural Networks for Computer Vision (ICLR 2024 · similarity 0.85)
- Deep Single Image Camera Calibration by Heatmap Regression to Recover Fisheye Images Under Manhattan World Assumption (CVPR 2024 · similarity 0.84)
- CAM Back Again: Large Kernel CNNs from a Weakly Supervised Object Localization Perspective (CVPR 2024 · similarity 0.84)
- Recurrent Partial Kernel Network for Efficient Optical Flow Estimation (AAAI 2024 · similarity 0.83)
- Implicit Zoo: A Large-Scale Dataset of Neural Implicit Functions for 2D Images and 3D Scenes (NeurIPS 2024 · similarity 0.83)
- Large Spatial Model: End-to-end Unposed Images to Semantic 3D (NeurIPS 2024 · similarity 0.83)
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.