Summary of Real Time Human Detection by Unmanned Aerial Vehicles, By Walid Guettala and Ali Sayah and Laid Kahloul and Ahmed Tibermacine
Real Time Human Detection by Unmanned Aerial Vehicles
by Walid Guettala, Ali Sayah, Laid Kahloul, Ahmed Tibermacine
First submitted to arxiv on: 6 Jan 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed UAV thermal infrared (TIR) object detection framework for images and videos leverages the Forward-looking Infrared (FLIR) cameras to generate a “You Only Look Once” (YOLO) model based on a convolutional neural network (CNN) architecture. This framework addresses the challenges of detecting small-scale objects in complex scenes with low resolution, limited publicly available labeled datasets, and training models. The YOLOv7 state-of-the-art model achieved an average precision of 72.5% at IOU=0.5 for human object detection during validation, while maintaining a detection speed of around 161 frames per second (FPS/second). This study demonstrates the effectiveness of the YOLO architecture in evaluating cross-detection performance of people in UAV TIR videos across varying observation angles. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Object detection from thermal infrared pictures and videos using deep-learning models is challenging due to limited publicly available labeled datasets, small-scale objects, complex scenes, and low resolution. This study proposes a framework for detecting objects in UAV TIR images and videos, which uses FLIR cameras to create a YOLO model based on CNN architecture. The results show that the YOLOv7 state-of-the-art model can detect human objects with 72.5% average precision at IOU=0.5 during validation, while maintaining a fast detection speed. |
Keywords
» Artificial intelligence » Cnn » Deep learning » Neural network » Object detection » Precision » Yolo
