Improving UAV Thermal Object Detection: A Comparative Assessment of Detection Approaches
Bibliographic record
Abstract
Object detection using Unmanned Aerial Vehicles (UAVs) has become increasingly vital for applications such as surveillance, traffic monitoring, and disaster response, particularly when utilizing infrared thermal imagery. However, detecting objects in high-altitude thermal images presents significant challenges, including low resolution, noise, complex environmental conditions, and varying object sizes. This study evaluates four state-of-the-art object detection models-You Only Look Once (YOLO) version 11, You Only Look Once with Neural Architecture Search (YOLONAS), You Only Look Once version 7 (YOLOv7), and Detection Transformer (DETR) on the HIT-UAV high-altitude infrared thermal dataset and the National Institute of Informatics Chiba University (NII-CU) multispectral aerial person detection dataset. The primary challenge addressed is the difficulty in detecting objects due to substantial noise, low resolution, and object overlapping in thermal images. A preprocessing process is proposed to enhance image quality, incorporating CLAHE (Contrast Limited Adaptive Histogram Equalization) for contrast improvement and bilateral filtering for noise reduction. This pipeline enhances the input images before they are fed into the detection models. Results demonstrate that YOLO-based models (YOLOv11, YOLO-NAS, and YOLOv7) outperform DETR, with YOLOv7 achieving the highest precision and recall. DETR underperformed significantly, highlighting the superior suitability of YOLO models for thermal UAV-based object detection in challenging environments. Detection was applied to both thermal video and image datasets, showcasing the efficacy of YOLO models over DETR in these scenarios.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.007 |
| Meta-epidemiology (narrow) | 0.002 | 0.000 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.004 | 0.002 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.002 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".