D
Debdoot Sheet
Researcher at Indian Institute of Technology Kharagpur
Publications - 129
Citations - 2813
Debdoot Sheet is an academic researcher from Indian Institute of Technology Kharagpur. The author has contributed to research in topics: Convolutional neural network & Segmentation. The author has an hindex of 19, co-authored 121 publications receiving 1824 citations. Previous affiliations of Debdoot Sheet include Jadavpur University & Ludwig Maximilian University of Munich.
Papers
More filters
Journal ArticleDOI
Designing Deep Neural High-Density Compression Engines for Radiology Images
Significance of Residual Learning and Boundary Weighted Loss in Ischaemic Stroke Lesion Segmentation
TL;DR: Radiologists use various imaging modalities to aid in different tasks like diagnosis of disease, lesion visualization, surgical planning and prognostic evaluation, and it was observed that addition of residual connections and boundary weighted loss improved the performance significantly.
Proceedings Article
Unit Impulse Response as an Explainer of Redundancy in a Deep Convolutional Neural Network
Rachana Sathish,Debdoot Sheet +1 more
TL;DR: A mechanism to empirically demonstrate the robustness in performance of aCNN on account of redundancy across its depth, a method to identify the systemic redundancy in response of a CNN across depth using the understanding of unit impulse response and use of these methods to interpret redundancy in few networks as example are proposed.
Patent
Multispectral optical imaging device and method for contactless functional imaging of skin
TL;DR: In this article, a multispectral optical imaging device, comprising of a driver circuit that triggers plurality of LEDs at different spectrum of illumination, a user-based platform/interface distributed over cloud that facilitates the communication between said device and a user; and a computational imaging server to process images according to server requirement and transfer the inferences back to device.
Posted Content
Identification of Cervical Pathology using Adversarial Neural Networks.
TL;DR: This paper proposes a convolutional autoencoder based framework, having an architecture similar to SegNet which is trained in an adversarial fashion for classifying images of the cervix acquired using a colposcope, and validate performance on the Intel-Mobile ODT cervical image classification dataset.