Research map / Detection and segmentation / CNN features
NU-AIR: A Neuromorphic Urban Aerial Dataset for Detection and Localization of Pedestrians and Vehicles
Craig Iaboni, Thomas Kelly, Pramod Abichandani
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 bias parameters (ablation) | VGG-11 + SSD SNN · NU-AIR test (SNN ablation) | COCO mAP: 0.11 -> 0.17 | Improved |
| add bias parameters (ablation) | DenseNet121-16 + SSD SNN · NU-AIR test (SNN ablation) | COCO mAP: 0.09 -> 0.17 | Improved |
| add bias parameters (ablation) | MobileNet-16 + SSD SNN · NU-AIR test (SNN ablation) | COCO mAP: 0.05 -> 0.13 | Improved |
| add dropout (0.5) after pooling layers (ablation) | VGG-11 + SSD SNN · NU-AIR test (SNN ablation) | COCO mAP: 0.17 -> 0.16 | Hurt |
| add dropout (0.5) after pooling layers (ablation) | MobileNet-16 + SSD SNN · NU-AIR test (SNN ablation) | COCO mAP: 0.13 -> 0.11 | Hurt |
| add dropout (0.5) after pooling layers (ablation) | DenseNet121 + SSD SNN · NU-AIR test (SNN ablation) | COCO mAP: 0.19 -> 0.16 | Hurt |
| use smaller SNN variants (MobileNet-13, VGG-9, DenseNet-98) instead of baseline depth (ablation) | SNN + SSD · NU-AIR test (SNN ablation) | COCO mAP: -8%, -6%, -6% relative mAP | Hurt |
| use larger SNN variants (MobileNet-19, VGG-16, DenseNet-142) (ablation) | SNN + SSD · NU-AIR test (SNN ablation) | COCO mAP: no gain over baseline | No clear effect |
| use Max Pooling instead of Average Pooling (ablation) | SNN + SSD · NU-AIR test (SNN ablation) | COCO mAP: Average Pooling: -6% (VGG-11), -16% (MobileNet-16), -12% (DenseNet121) | Improved |
| use BatchNorm instead of LayerNorm (ablation) | SNN + SSD · NU-AIR test (SNN ablation) | COCO mAP: LayerNorm: -6% (DenseNet121, VGG-11), -16% (MobileNet-16) | Improved |
| event-based SNN instead of frame-based DNN detectors | object detection · NU-AIR test | COCO mAP: 0.41 -> 0.16 | Hurt |
| use YOLOv7 instead of YOLOv7-tiny | YOLO detector · NU-AIR test | COCO mAP: 0.36 -> 0.53 | Improved |
Closest papers by meaning
- MatrixCity: A Large-scale City Dataset for City-scale Neural Rendering and Beyond (ICCV 2023 · similarity 0.85)
- SUDS: Scalable Urban Dynamic Scenes (CVPR 2023 · similarity 0.84)
- UrbanIng-V2X: A Large-Scale Multi-Vehicle, Multi-Infrastructure Dataset Across Multiple Intersections for Cooperative Perception (NeurIPS 2025 · similarity 0.83)
- SFOD: Spiking Fusion Object Detector (CVPR 2024 · similarity 0.83)
- Swift-Mapping: Online Neural Implicit Dense Mapping in Urban Scenes (AAAI 2024 · similarity 0.83)
- Efficient Event-Based Object Detection: A Hybrid Neural Network with Spatial and Temporal Attention (CVPR 2025 · 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.
