Win in Data Mining Hackathon!
After a top-three result last year, Hien Dang won this year's 2nd Data Mining Hackathon at the 21st IEEE International Conference on Intelligent Transportation Systems. Congratulations!
During 10 hours of the hackathon, the participants were asked to develop a model for predicting lane changes on highway scenario. The provided data consists of video from a forward-facing camera and IMU sensor data (Inertial Measurement Unit).
The winning solution of Hien Dang uses a deep neural network, that captures the video data with convolutional layers and the IMU data with fully connected layers. The architecture then fuses inputs from both video and IMU data to predict lane change.