Research map / Detection and segmentation / CNN features
Per‑Object IoU Forecasting for Deadline‑Aware Real‑Time Embedded Detection Control
Erfan Foorginejad, Akshar Chavan, Marco Brocanelli
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 |
|---|---|---|---|
| model deadlines per object instead of frame-level average (OLAP-tiny vs FLAP) (ablation) | deadline predictor · Deadline prediction | RMSE: 32.3% average RMSE reduction | Improved |
| add object dynamics features (speed, location, size) to OLAP-tiny (ablation) | OLAP · Deadline prediction | RMSE: 34.5% additional RMSE reduction | Improved |
| use OLAP deadlines in controller instead of ROMA | OLAP-C · Streaming detection, Argoverse-HD + MOT17 (71 scenarios), Intel Node | sAP: +39% over ROMA (0.287) | Improved |
| use OLAP-C instead of ROMA on Apple M3 | OLAP-C · Streaming detection, Argoverse-HD + MOT17 (71 scenarios), Apple M3 | sAP: +24.6% over ROMA | Improved |
| use OLAP-C instead of ROMA on Jetson | OLAP-C · Streaming detection, Argoverse-HD + MOT17 (71 scenarios), Jetson | sAP: +1.3% over ROMA (0.408) | No clear effect |
| adaptive detector selection instead of best static detector | ROMA · Streaming detection, Argoverse-HD + MOT17 (71 scenarios), Intel Node | sAP: +25% over Best Static | Improved |
| OLAP-C vs frame-level FLAP for longer deadlines (>15 frames) | OLAP · Deadline prediction | error: up to 67% and 56% average improvement but conservative | Mixed |
| FLAP on short deadlines (up to 6 frames) | FLAP · Deadline prediction | deadline accuracy | No clear effect |
| raise priority weight of critical object to 10 (ablation) | OLAP-C · MOT17-13 pedestrian example | per-object IoU: critical object IoU up, non-critical average IoU down | Mixed |
Closest papers by meaning
- Can't Slow Me Down: Learning Robust and Hardware-Adaptive Object Detectors against Latency Attacks for Edge Devices (CVPR 2025 · similarity 0.84)
- Object Detection using Event Camera: A MoE Heat Conduction based Detector and A New Benchmark Dataset (CVPR 2025 · similarity 0.82)
- Transtreaming: Adaptive Delay-aware Transformer for Real-time Streaming Perception (AAAI 2025 · similarity 0.81)
- YOLOv10: Real-Time End-to-End Object Detection (NeurIPS 2024 · similarity 0.81)
- DEIM: DETR with Improved Matching for Fast Convergence (CVPR 2025 · similarity 0.81)
- Ev-3DOD: Pushing the Temporal Boundaries of 3D Object Detection with Event Cameras (CVPR 2025 · 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.
