Kazuma Yamamoto
Papers
1
Total Citations
55
H-Index
1
About
Kazuma Yamamoto has made significant contributions to computer vision, particularly in object detection and deep learning methodologies. His most cited work, "Resolving Class Imbalance in Object Detection with Weighted Cross Entropy Losses" (2020, 55 citations), addresses a critical challenge in real-world vision systems: the imbalance between foreground and background classes during training. By introducing weighted cross entropy loss functions, Yamamoto provided a practical solution that improves detection accuracy in applications ranging from autonomous driving to surveillance and robotics. This work has been influential in advancing generalized object detectors like Faster R-CNN, YOLO, and SSD. Yamamoto’s research bridges the gap between theoretical loss optimization and robust, deployable vision systems. His contributions are particularly valued in scenarios where class imbalance degrades model performance, making his methods essential for practitioners building reliable AI systems. With a growing citation impact, Yamamoto continues to shape the development of more equitable and effective deep learning architectures for visual recognition tasks.
Research Focus
Key Achievements
Top Papers
- 1