Research map / Detection and segmentation / CNN features
EvRT-DETR: Latent Space Adaptation of Image Detectors for Event-based Vision
Dmitrii Torbunov, Yihui Ren, Animesh Ghose, Odera Dim, Yonggang Cui
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 frozen-detector temporal ConvLSTM adaptation (I2EvDet) to RT-DETR-B | RT-DETR-B · Gen1 test | mAP (%): 47.6 -> 52.7 | Improved |
| add temporal adaptation to RT-DETR-T | RT-DETR-T · Gen1 test | mAP (%): 46.0 -> 52.3 | Improved |
| add temporal adaptation to RT-DETR-B | RT-DETR-B · 1Mpx/Gen4 test | mAP (%): 45.2 -> 50.1 | Improved |
| add temporal adaptation to YOLOX-X | YOLOX-X · Gen1 test (YOLOX) | mAP (%): 43.4 -> 47.8 | Improved |
| add temporal adaptation to YOLOX-T | YOLOX-T · Gen1 test (YOLOX) | mAP (%): 36.0 -> 42.4 | Improved |
| add temporal module at lowest-level scale E3 (ablation) | EvRT-DETR-B · Gen1 test | mAP (%): 51.0 -> 52.7 | Improved |
| add temporal module at scale E4 (ablation) | EvRT-DETR-B · Gen1 test | mAP (%): 52.2 -> 52.7 | Improved |
| add temporal module at scale E5 (ablation) | EvRT-DETR-B · Gen1 test | mAP (%): 52.4 -> 52.7 | Improved |
| reduce ConvLSTM hidden dim from 256 to 64 (ablation) | EvRT-DETR-B · Gen1 test | mAP (%): 52.7 -> 52.1 | Hurt |
| increase ConvLSTM hidden dim from 256 to 512 (ablation) | EvRT-DETR-B · Gen1 test | mAP (%): 52.7 -> 52.9 | No clear effect |
| use 3x3 ConvLSTM kernel instead of 1x1 (ablation) | EvRT-DETR-B · Gen1 test | mAP (%): 52.3 -> 52.7 | Improved |
| use 3x3 kernel instead of 5x5 (ablation) | EvRT-DETR-B · Gen1 test | mAP (%): 52.0 -> 52.7 | Improved |
Closest papers by meaning
- FlexEvent: Towards Flexible Event-Frame Object Detection at Varying Operational Frequencies (NeurIPS 2025 · similarity 0.88)
- EGSST: Event-based Graph Spatiotemporal Sensitive Transformer for Object Detection (NeurIPS 2024 · similarity 0.88)
- Context Enhanced Transformer for Single Image Object Detection in Video Data (AAAI 2024 · similarity 0.87)
- PASS: Path-selective State Space Model for Event-based Recognition (NeurIPS 2025 · similarity 0.86)
- E2HQV: High-Quality Video Generation from Event Camera via Theory-Inspired Model-Aided Deep Learning (AAAI 2024 · similarity 0.86)
- Scene Adaptive Sparse Transformer for Event-based Object Detection (CVPR 2024 · similarity 0.85)
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.
