Title: Energy-aware automatic video annotation tool for autonomous vehicle

Authors: N.S. Manikandan; K. Ganesan

Addresses: TIFAC-CORE, Vellore Institute of Technology (VIT), Vellore 632014, India ' School of Information Technology and Engineering, VIT Business School, Vellore Institute of Technology (VIT), Vellore 632014, India

Abstract: In a self-driving car, real-time video obtained from the camera sensors is analysed using various scene understanding algorithmic modules (object detection, object classification, lane detection and object tracking). In this paper, we propose an annotation tool that uses deep learning techniques for each of the four modules mentioned above, and the best ones are chosen based on suitable metrics. Our tool is 83% accurate when compared with a human annotator. We considered a video with 530 frames of resolution 1,035 × 1,800 pixels. Our proposed tool consumed 43 minutes of computation with 36.73 g of CO₂ emission in a CPU-based system and 2.58 minutes of computation with 7.75 g of CO₂ emission in a GPU-based system to process all four modules. But the same video took nearly 3,060 minutes of computational usage with 2.56 kg of CO₂ emission for one human annotator to narrate the scene using a normal computer.

Keywords: automatic annotation; deep learning; object classification; object detection; lane detection; object tracking.

DOI: 10.1504/IJCVR.2023.133137

International Journal of Computational Vision and Robotics, 2023 Vol.13 No.5, pp.510 - 532

Received: 09 Nov 2021
Accepted: 08 May 2022
Published online: 01 Sep 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article