On-the-road vehicle and pedestrian detection for self-driving cars thus far has been a combination of sonar, lidar, and cameras. For cameras, although object detection has improved dramatically in the last 20 years, common algorithms are best suited to handle static images and treat videos as a series of disjoint detections. Sequences across time can hold extra information allowing for improved detection performance as well as the added benefit of tracking objects through time. Thsi research doubles a standard fully convolutional network into a siamese architecture to handle consecutive frame inputs and adds a tracking module to assist in the prediction of frame to frame differences. The architecture is evaluated on both a standard dataset and a custom dataset to test its ability to handle more in-scene complexity.
Combined Detection and Tracking Impacts on Vehicle Specific Data
Researchers may make free and open use of the UC Berkeley Library’s digitized public domain materials. However, some materials in our online collections may be protected by U.S. copyright law (Title 17, U.S.C.). Use or reproduction of materials protected by copyright beyond that allowed by fair use (Title 17, U.S.C. § 107) requires permission from the copyright owners. The use or reproduction of some materials may also be restricted by terms of University of California gift or purchase agreements, privacy and publicity rights, or trademark law. Responsibility for determining rights status and permissibility of any use or reproduction rests exclusively with the researcher. To learn more or make inquiries, please see our permissions policies (https://www.lib.berkeley.edu/about/permissions-policies).