Research map / Video / Video segmentation and tracking
BDD100K: A Diverse Driving Dataset for Heterogeneous Multitask Learning
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 |
|---|---|---|---|
| Add object detection supervision | BDD100K instance segmentation · BDD100K instance segmentation | AP: 21.8 -> 24.5 | Helped |
| Add object detection supervision | BDD100K MOT · BDD100K MOT | MOTA: 55.0 -> 56.7 | Helped |
| Add object detection supervision | BDD100K MOT · BDD100K MOT | IDS: 8386 -> 9098 | Hurt |
| Add object detection supervision | BDD100K semantic segmentation · BDD100K semantic segmentation | mean IoU: 56.9 -> 58.3 | Helped |
| Add lane marking and drivable-area supervision | BDD100K semantic segmentation · BDD100K semantic segmentation | mean IoU: 56.9 -> 53.3 | Hurt |
| Fine-tune from instance segmentation | BDD100K MOTS · BDD100K MOTS | AP: 13.0 -> 18.7 | Helped |
| Fine-tune from MOT | BDD100K MOTS · BDD100K MOTS | MOTSA: 30.4 -> 40.3 | Helped |
| Jointly train detection, tracking, instance segmentation, and MOTS | BDD100K MOTS · BDD100K MOTS | MOTSA: 30.4 -> 41.4 | Helped |
| Add drivable-area supervision | BDD100K lane marking with 10K training images · BDD100K lane marking, 10K training images | mean ODS-F (τ=10): 45.41 -> 50.40 | Helped |
| Add drivable-area supervision | BDD100K lane marking with 20K training images · BDD100K lane marking, 20K training images | mean ODS-F (τ=10): 54.36 -> 54.11 | No clear effect |
| Add drivable-area supervision | BDD100K lane marking with 70K training images · BDD100K lane marking, 70K training images | mean ODS-F (τ=10): 54.48 -> 54.24 | No clear effect |
| Add lane-marking supervision | BDD100K drivable area with 10K training images · BDD100K drivable area, 10K training images | mean IoU: 64.23 -> 64.37 | No clear effect |
Closest papers by meaning
- Visual Exemplar Driven Task-Prompting for Unified Perception in Autonomous Driving (CVPR 2023 · similarity 0.89)
- One Million Scenes for Autonomous Driving: ONCE Dataset (NeurIPS 2021 · similarity 0.88)
- M3Net: Multimodal Multi-task Learning for 3D Detection, Segmentation, and Occupancy Prediction in Autonomous Driving (AAAI 2025 · similarity 0.86)
- The Road Ahead in Autonomous Driving: The KITScenes Multimodal Dataset (NeurIPS 2026 · similarity 0.85)
- Generalized Predictive Model for Autonomous Driving (CVPR 2024 · similarity 0.85)
- DriveDreamer-2: LLM-Enhanced World Models for Diverse Driving Video Generation (AAAI 2025 · 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.