Article
AERIAL IMAGE SEMANTIC SEGMENTATION USING U-NET AND EFFICIENTNET
To address the challenges of few-shot aerial image semantic segmentation, where unseen-category objects in query aerial images need to be parsed with only a few annotated support images, we propose a novel approach by integrating a U-Net architecture with Efficient Net. Typically, in few-shot segmentation, category prototypes are extracted from support samples to segment query images in a pixel-wise matching process. However, the arbitrary orientations and distribution of aerial objects in such images often result in significant feature variations, making conventional methods, which do not account for orientation changes, ineffective. The rotation sensitivity of aerial images causes substantial feature distortions, leading to low confidence scores and misclassification of same-category objects with different orientations. To overcome these limitations, we propose an enhanced solution by combining U-Net for robust semantic segmentation with Efficient Net for efficient feature extraction and scale adaptability. This architecture, which we refer to as Efficient U-Net, introduces rotation-invariant feature extraction to handle the varying orientations of aerial objects. By leveraging Efficient Net’s scalable Convolutional layers for feature extraction, we ensure that the network can capture orientation-varying yet category-consistent information from support images. This approach enhances the segmentation accuracy by aligning same-category objects, irrespective of their orientation, thereby minimizing the oscillation of confidence scores and improving the detection of rotated semantic objects. This Efficient U-Net model provides a scalable, rotation-invariant solution to the few-shot segmentation of aerial images.
Full Text Attachment
