TY  - CONF
AU  - Lorenzo Porzi
AU  - Markus Hofinger
AU  - Idoia Ruiz
AU  - Joan Serrat
AU  - Samuel Rota Bulo
AU  - Peter Kontschieder
A2  - CVPR
PY  - 2020//
TI  - Learning Multi-Object Tracking and Segmentation from Automatic Annotations
BT  - 33rd IEEE Conference on Computer Vision and Pattern Recognition
SP  - 6845
EP  - 6854
N2  - In this work we contribute a novel pipeline to automatically generate training data, and to improve over state-of-the-art multi-object tracking and segmentation (MOTS) methods. Our proposed track mining algorithm turns raw street-level videos into high-fidelity MOTS training data, is scalable and overcomes the need of expensive and time-consuming manual annotation approaches. We leverage state-of-the-art instance segmentation results in combination with optical flow predictions, also trained on automatically harvested training data. Our second major contribution is MOTSNet – a deep learning, tracking-by-detection architecture for MOTS – deploying a novel mask-pooling layer for improved object association over time. Training MOTSNet with our automatically extracted data leads to significantly improved sMOTSA scores on the novel KITTI MOTS dataset (+1.9%/+7.5% on cars/pedestrians), and MOTSNet improves by +4.1% over previously best methods on the MOTSChallenge dataset. Our most impressive finding is that we can improve over previous best-performing works, even in complete absence of manually annotated MOTS training data.
UR  - https://ieeexplore.ieee.org/abstract/document/9157138
L1  - http://refbase.cvc.uab.es/files/PHR2020.pdf
UR  - http://dx.doi.org/10.1109/cvpr42600.2020.00688
N1  - ADAS; 600.124; 600.118
ID  - Lorenzo Porzi2020
ER  -