Improving Cattle Detection in Low Light with Adaptive Image Enhancement
Author Information
Author(s): Zheng Zhiqiang, Wang Mengbo, Zhao Xiaoyu, Weng Zhi
Primary Institution: Inner Mongolia University
Hypothesis
Can adaptive image enhancement improve cattle detection accuracy in non-ideal lighting conditions?
Conclusion
The AT-Detr algorithm significantly enhances cattle detection accuracy in low-light conditions, achieving 97.5% accuracy.
Supporting Evidence
- The AT-Detr algorithm achieved a recall of 99.4%, reflecting a 1% improvement over previous methods.
- Image enhancement using the Adltformer model improved PSNR and SSIM metrics significantly.
- The study utilized a dataset of 3376 images captured under various lighting conditions.
Takeaway
This study shows that using special techniques to improve images can help computers find cows better, even when it's dark.
Methodology
The study used a combination of adaptive image enhancement and team-training with a detection model to improve cattle detection in low-light conditions.
Potential Biases
Potential biases may arise from the dataset used, which may not fully represent all lighting conditions.
Limitations
The study primarily focused on low-light conditions and may not generalize to other environmental factors affecting detection.
Participant Demographics
Cattle images were collected from various locations under different lighting conditions.
Statistical Information
P-Value
0.0001
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website