Visual Intelligence & Logistics
2022-05-30 – 2022-11-30 · Completed
Korea Railroad Research Institute (KRRI)
Project titles
- Development of a Cargo Information Recognition Prototype (화물 정보 인식기 시제품 제작)
- Construction of a Training Database for Deep Learning-Based Cargo Damage Recognition (딥러닝 기반의 화물 파손 인식을 위한 학습 DB 구축)
Research
Integrated RGB-camera and LiDAR perception with loading-decision interfaces and built point-cloud data for cargo and container space.
Results
Certified test results reported in 2022.
| Task | Result | Test conditions |
|---|---|---|
| Position and volume perception | mIoU 0.867 | Ten unseen normal cargo types · 100 trials |
| Cargo classification and barcode | 100% joint success | Both functions had to succeed |
| Damage recognition | 94% success | Normal and damaged cargo groups · 100 trials |