Few-shot semantic segmentation aims to segment novel-class objects in a query image with only a few annotated examples in support images. Although progress has been made recently by combining prototype-based metric learning, existing methods still face two main challenges. First, various intra-class objects between the support and query images or semantically similar inter-class objects can seriously harm the segmentation performance due to their poor feature representations. Second, the latent novel classes are treated as the background in most methods, leading to a learning bias, whereby these novel classes are difficult to correctly segment as foreground. To solve these problems, we propose a dual-branch learning method. The class-specific branch encourages representations of objects to be more distinguishable by increasing the inter-class distance while decreasing the intra-class distance. In parallel, the class-agnostic branch focuses on minimizing the foreground class feature distribution and maximizing the features between the foreground and background, thus increasing the generalizability to novel classes in the test stage. Furthermore, to obtain more representative features, pixel-level and prototype-level semantic learning are both involved in the two branches. The method is evaluated on PASCAL-5i1-shot, PASCAL-5i5-shot, COCO-20i1-shot, and COCO-20i5-shot, and extensive experiments show that our approach is effective for few-shot semantic segmentation despite its simplicity.

For seismically thin-reservoir layers, variations in rock properties may not be directly linked to seismic amplitude due to the wave interference of layer top and base reflections. In addition, thin-layer reflection signal locally has a different phase from that of the signal wavelet. Signal even and odd components can be considered as amplitudes at different signal phases, which may have a different sensitivity to the variations in thin layer and surrounding layer properties. A novel extension of the spectral decomposition concept is proposed that decomposes seismic signal into its even and odd components via the short-time Fourier transform. Amplitude attributes for the original signal and even and odd part components are compared for their ability to restore the correct “amplitude-layer property” correlation without resolving the thin layer. Numerical modeling analysis shows that amplitude at peak frequency (APF) of the seismic data odd component APF (OAPF) is more sensitive to thin-reservoir property change compared to the conventional APF and even component APF attributes. When applied in analyzing real seismic data in a tight-dolomite reservoir, conventional APF and conventional acoustic impedance inversion did not provide a correct relationship to porosity variations. Meanwhile, the OAPF attribute responds well to porosity measured in boreholes. This suggests that the interpretability of amplitude attributes in thin layers can be improved by signal even and odd decomposition.