Autonomous Driving from an inboard camera is a highly complex task that can be approached using computer vision techniques. In this paper we present a deep learning approach for scene understanding for autonomous vehicles, focusing on 3 stages of increasing complexity: object recognition, object detection and semantic segmentation of the scene. This three stages have a common goal, understand the scene and make consequent navigation decisions. For the object recognition stage we have tested several Convolutional Neural Networks in a variety of datasets to asses their performance.
- Arnau Baró - [email protected]
- Pau Cebrián - [email protected]
- Victor Campmany - [email protected]
- Guillem Cucurull - [email protected]
The paper will be updated weekly according to the work done in the project.
Scene Understanding for Autonomous Vehicles
Slides summarizing the work done each week
Explanations of the work and tasks done each week
Simonyan, K., & Zisserman, A. (2014). Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556. Paper summary
He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 770-778). Paper summary
Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J., & Wojna, Z. (2016). Rethinking the inception architecture for computer vision. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 2818-2826). Paper summary