Domain Adaptation for Pedestrian DCNN Detector toward a Specific Scene and an Embedded Platform.
Nada HammamiAla MhallaAlexis LandraultPublished in: VISIGRAPP (5: VISAPP) (2019)
Keyphrases
- domain adaptation
- cross domain
- multiple sources
- pedestrian detector
- domain specific
- manifold alignment
- labeled data
- semi supervised
- data sets
- transfer learning
- document classification
- semi supervised learning
- training data
- image sequences
- three dimensional
- unlabeled data
- test data
- input image
- co training
- covariate shift