TiledSoilingNet: Tile-level Soiling Detection on Automotive Surround-view Cameras Using Coverage Metric

Abstract

Automotive cameras, particularly surround-view cameras, tend to get soiled by mud, water, snow, etc. For higher levels of autonomous driving, it is necessary to have a soiling detection algorithm which will trigger an automatic cleaning system. Localized detection of soiling in an image is necessary to control the cleaning system. It is also necessary to enable partial functionality in unsoiled areas while reducing confidence in soiled areas. Although this can be solved using a semantic segmentation task, we explore a more efficient solution targeting deployment in low power embedded system. We propose a novel method to regress the area of each soiling type within a tile directly. We refer to this as coverage. The proposed approach is better than learning the dominant class in a tile as multiple soiling types occur within a tile commonly. It also has the advantage of dealing with coarse polygon annotation, which will cause the segmentation task. The proposed soiling coverage decoder is an order of magnitude faster than an equivalent segmentation decoder. We also integrated it into an object detection and semantic segmentation multi-task model using an asynchronous back-propagation algorithm. A portion of the dataset used will be released publicly as part of our WoodScape dataset [1] to encourage further research.

Multi-Task Network Pruning and Embedded Optimization for Real-time Deployment in ADAS

“…Soiling detection Soiling detection [31] can be approached as a coarse semantic segmentation task, based on a 4 × 4 grid for localization as described in [8].…”

“…Particularly the network depth, the number of filters and the position of skip connections and other layers have been optimized to boosted performance while limiting the network complexity. We used a YOLO-like decoder for object detection, a FCN-like decoder for semantic segmentation and for soiling we used the decoder described in [8]. The kernel size was set to 5 × 5, which has an efficient implementation on the used embedded system.…”