Santosh Kumar Vipparthi

Micro-expressions (MEs) are transient in nature and reveal enough visual cues to recognize genuine emotions. Neural Architecture Search (NAS) has recently gained significant attention in the field of micro expression recognition (MER). However, existing NAS approaches in MER rely on general-purpose strategies, often reducing network depth to compensate for limited training data. This restriction can limit the discovery of optimal architectures, especially since dataset sizes in MER vary significantly. Moreover, focusing solely on operation selection and connectivity is inadequate, flexibility in depth and structure is essential to adapt to different data characteristics. While shallow networks are often considered suitable for small datasets, this assumption falls short for MER. Despite the scarcity of training samples, MEs are inherently complex, fleeting, and nuanced, requiring models with high representational power for accurate recognition. To address this trade-off, we propose a novel Dynamic Neural Architecture Search framework for MER (DNAS-MER), designed to adaptively balance model depth and architectural design based on the data. The proposed DNAS hierarchically optimizes the network depth, resolution paths, and cell-level operations through a scalable network search space and a multi-scale context-aware (MSCA) cell search space. A novel depth-aware loss function enables the model to automatically adapt depth based on data needs, while MSCA cell search, embedded with adaptive feature fusion operations, enhances ME-specific feature representation to better capture the complexity of MEs. We evaluate DNAS-MER on a composite dataset MEGC2019, CASME-II, and SMIC. Experimental results demonstrate that DNAS-MER consistently outperforms existing state-of-the-art methods for both video and image-based MER tasks.

Underwater image restoration aims to improve the quality and visibility of images taken in underwater environments. These images find application in diverse fields like marine biology research, underwater archaeology, environmental monitoring, surveillance tasks, and offshore infrastructure inspection. However, the complexities of the underwater environment make these applications challenging, as light scattering and absorption cause blur, color cast, and reduced contrast in images. With the promising results on restoring underwater degraded images, existing approaches limit their performance in case of the above-mentioned complex and nonlinear degradation. In this research work, we propose a multi-directional wavelet coefficient space transformer model for underwater image deblurring and color restoration. Incorporating an attention mechanism within transformed spaces, our model dynamically adapts to underwater degradation. Additionally, we introduce a wavelet attention fusion transformer block for attention computation in the wavelet coefficient space, along with an edge-preserving wavelet down-sampling block to retain fine details and textures during downsampling. A thorough assessment of our method on real-world (UCCS, U45, SQUID) and synthetic (UIEB, UCDD) datasets, along with profound ablation studies, validates its edge over existing techniques. Further, we have evaluated our method for tasks such as depth estimation, and low-light enhancement and deblurring, demonstrating its versatility and broad applicability across various image processing tasks. The code is made available at: https://github.com/Priyanka01mishra/TransWaveNet.