Improving Object Detection with Super-Resolution Technology
Author Information
Author(s): Marta BistroĊ, Zbigniew Piotrowski
Primary Institution: Military University of Technology, Faculty of Electronics
Hypothesis
Can super-resolution technology enhance the performance of image reconnaissance systems under interference conditions?
Conclusion
The study found that super-resolution significantly improved detection precision and mean average precision in most interference scenarios.
Supporting Evidence
- Super-resolution improved detection precision in most scenarios.
- The Faster R-CNN model was used for object detection.
- Motion blur significantly reduced detection performance.
- Super-resolution was effective in enhancing image quality.
- Training with high-resolution data improved model accuracy.
Takeaway
This study shows that using special technology can make blurry pictures clearer, helping computers find objects better, especially when things are not perfect.
Methodology
The study involved training and evaluating the Faster R-CNN detection model with original and modified datasets under various interference conditions.
Limitations
The super-resolution model struggled with motion blur and complex combinations of distortions, which affected detection performance.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website