ShapeEmbed: a self-supervised learning framework for 2D contour quantification
This work addresses shape quantification challenges in applications like biological imaging, offering an incremental improvement over existing autoencoder-based methods.
The paper tackled the problem of quantifying 2D object shapes with invariance to geometric transformations, introducing ShapeEmbed, a self-supervised learning framework that encodes contours into shape descriptors, which outperformed competitors in shape classification tasks on natural and biological images.
The shape of objects is an important source of visual information in a wide range of applications. One of the core challenges of shape quantification is to ensure that the extracted measurements remain invariant to transformations that preserve an object's intrinsic geometry, such as changing its size, orientation, and position in the image. In this work, we introduce ShapeEmbed, a self-supervised representation learning framework designed to encode the contour of objects in 2D images, represented as a Euclidean distance matrix, into a shape descriptor that is invariant to translation, scaling, rotation, reflection, and point indexing. Our approach overcomes the limitations of traditional shape descriptors while improving upon existing state-of-the-art autoencoder-based approaches. We demonstrate that the descriptors learned by our framework outperform their competitors in shape classification tasks on natural and biological images. We envision our approach to be of particular relevance to biological imaging applications.