Abstract: Existing image-based fire segmentation techniques use convolutional neural networks to handle complicated scenes. Such approaches perform poorly when flame sizes vary greatly and when ...
Abstract: Oriented object detection in aerial images has made significant advancements propelled by well-developed detection frameworks and diverse representation approaches to oriented bounding boxes ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback