BROWSE
Detail
Conference Paper
딥러닝 기반의 차량 후미등 신호 분류기 연구
Cited - time in 
Share 
Share
- Issue Date
- 2019-11
- Citation
- 한국자동차공학회 학술 대회 (추계) 2019, pp.871-874
- Publisher
- 한국자동차공학회
- Language
- Korean
- Type
- Conference Paper
- Abstract
- Classification of taillight signal for forward vehicles is an important clue for self-driving plans. Accurate taillight recognition requires temporal context grasp over several frames. We build a new dataset that accurately reflects the situation where the taillight states change over time. Using the dataset, We train and evaluate two deep learning models that can learn temporal context. First model is the sequence to sequence network which uses RNN, and second model is the C3D network which uses 3D CNN.
- KSP Keywords
- 3D CNN, Over time, Self-Driving, Sequence network, Sequence to sequence, deep learning(DL), deep learning models, temporal context