Abstract:
Imaging Systems have progressed a lot in the last few decades. With the advent of Machine Learning (ML) and Artificial Intelligence (AI), a new paradigm of imaging systems has evolved. In this talk, I will present my work on imaging systems driven by AI for material characterization. I would outline my works including - (i) task specific intelligent scene acquisition, (ii) neuromorphic computing algorithms for object recognition and tactile sensing, (iii) physics-based machine learning for radiation sensor characterization, and (iv) developing computational imaging pipeline for 3D imaging of magnetic domains at nanoscale. Subsequently, I would present in detail - (i) a novel physics-based machine learning model to infer material properties at microscopic level for the RTSD. This approach proposes a trainable model based on physical charge transport with trapping centres for electrons and holes in the detector. The RTSD is divided spatially into voxels, and in each voxel the different material properties are modeled as trainable weights which are trained with signals at the electrodes and free and trapped charges in the material, and (ii) a Scanning Transmission X-ray Microscopy combined with Vector Computed Tomography approach using Hard X-rays to characterize the 3D magnetic domains at nanoscale. I would conclude my talk with future research plans both in short and long term.