DeepTrack

Lightweight Deep Learning for Vehicle Trajectory Prediction

DeepTrack presents a lightweight deep learning approach for vehicle trajectory prediction in highways using Temporal Convolutional Networks (TCNs).

Key Features

  • Depthwise separable convolutions for reduced model complexity
  • State-of-the-art accuracy with significantly reduced computational requirements
  • Designed for IoT deployment in intelligent transportation systems
  • 33.16% reduction in model size compared to LSTM-based approaches

Publication

V. Katariya, M. Baharani, et al., “DeepTrack: Lightweight Deep Learning for Vehicle Trajectory Prediction in Highways,” IEEE Transactions on Intelligent Transportation Systems, vol. 23, no. 10, pp. 18927-18936, 2022. (97 citations)

Paper Code