Loading...
Please wait, while we are loading the content...
Similar Documents
Deep Drone : Object Detection and Tracking for Smart Drones on Embedded System
| Content Provider | Semantic Scholar |
|---|---|
| Author | Han, Song Shen, William Liu, Zuozhen |
| Copyright Year | 2016 |
| Abstract | In recent years, drones have been widely adopted for aerial photography at much lower costs. However, capturing high quality pictures or videos using most advanced drones requires precise manual control and are very errorprone. We are proposing Deep Drone, an embedded system framework, to power drones with vision: letting the drone to do automatic detection and tracking. In this project, we implemented the vision component which is an integration of advanced detection and tracking algorithms. We implemented our system onto multiple hardware platforms, including both desktop GPU (NVIDIA GTX980) and embedded GPU (NVIDIA Tegra K1 and NVIDIA Tegra X1) and evaluated frame rate, power consumption and accuracy on several videos captured by the drone. Our system achieved real time performance at 71 frames per second(fps) for tracking and 1.6 fps for detection on NVIDIA TX1. The video demo of our detection and tracking algorithm has been uploaded to Youtube: https://youtu. be/UTx2-5a488s. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | https://web.stanford.edu/class/cs231a/prev_projects_2016/deep-drone-object__2_.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
