A novel target-oriented enhanced infrared camera trap data screening method DOI Creative Commons

Yanfei Cai,

Kaikai Tian,

Ji Liang

et al.

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: May 10, 2025

Infrared Camera Traps (ICTs) are widely used in ecological research as a noninvasive wildlife monitoring technique, particularly for the detection and identification of animal targets. Existing ICT data screening methods face challenges recognizing animals against complex backgrounds, fast-moving or small To address these issues, we proposed target-oriented enhanced data-screening method called GFD-YOLO, which emphasized key locations images to effectively guide focus model toward target regions, thereby improving accuracy. We compared effects different preprocessing on performance. Results revealed that improved mean Average Precision (mAP) by 16.96%, precision 10.13%, recall 24.85% YOLOv11n model. Therefore, this study had significant advantages reducing false negatives positives was adaptable tasks under background conditions. In addition, demonstrated higher robustness scenarios involving lighting variations

Language: Английский

Thinking coexistence in human-dominated landscapes with the lens of multi-species assemblages: Farmers, brown bears and other wild species in the Pyrenees DOI

Manon Culos,

Alice Ouvrier,

Ruppert Vimal

et al.

Biological Conservation, Journal Year: 2025, Volume and Issue: 302, P. 111006 - 111006

Published: Feb. 1, 2025

Language: Английский

Citations

0
A novel target-oriented enhanced infrared camera trap data screening method DOI Creative Commons

Yanfei Cai,

Kaikai Tian,

Ji Liang

et al.

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: May 10, 2025

Infrared Camera Traps (ICTs) are widely used in ecological research as a noninvasive wildlife monitoring technique, particularly for the detection and identification of animal targets. Existing ICT data screening methods face challenges recognizing animals against complex backgrounds, fast-moving or small To address these issues, we proposed target-oriented enhanced data-screening method called GFD-YOLO, which emphasized key locations images to effectively guide focus model toward target regions, thereby improving accuracy. We compared effects different preprocessing on performance. Results revealed that improved mean Average Precision (mAP) by 16.96%, precision 10.13%, recall 24.85% YOLOv11n model. Therefore, this study had significant advantages reducing false negatives positives was adaptable tasks under background conditions. In addition, demonstrated higher robustness scenarios involving lighting variations

Language: Английский

Citations

0