Patents Filed by Dr. Mrinal Kanti Bhowmik

A system and method for segmenting suspicious hyperthermic regions from breast thermograms

New Indian Patent Application No: 202031027352

Status: Published Old PDF New PDF

Abstract: The Appearance of suspicious hyperthermic regions (SHRs) in breast thermograms is the single most marker of breast abnormality. Hence, accurate segmentation and analysis of SHRs are very crucial for grading the degree of severity in breast thermograms. A novel breast abnormality grading approach namely Morphology Model of Suspicious Hyperthermic Regions (MMSHRs) has been proposed here. The proposed MMSHRs method first segments the SHRs and then, the morphology of the SHRs has been analyzed to grade the thermograms according to their degree of severity. The experimental results show that the proposed segmentation method can extract the SHRs more accurately with higher average accuracy rate compared to the other state-of-the-art methods. The segmentation of SHRs is followed by the extraction of morphological features of SHRs, which categorizes the abnormal thermograms into mild abnormal and severely abnormal with classification accuracy of 91% based on the degree of severity present in the thermograms.

Type: Application

Filed: June 27, 2020

Publication Date: December 31, 2021

Applicants: Usha Rani Gogoi (DST - INSPIRE Fellow, Department of Computer Science and Engineering, Tripura University), Dr. Mrinal Kanti Bhowmik (Assistant Professor, Department of Computer Science & Engineering, Tripura University)

System and Method for Detecting Object in Adverse Atmosphere by Restoring Degraded Image in Deep Convolutional Layer

New Indian Patent Application No: 202131002651

Status: Published Old PDF New PDF

Abstract: The present invention discloses a system and method for detecting objects in real-time adverse weather-degraded scenes. A single-stage CNN architecture is adopted, namely, AWDRDNet for detecting objects more accurately in adverse weather-degraded realistic scenes. The present invention relates to a feed-forward deeper convolutional layer comprising a plurality of convolution blocks (B1,..,BK,..,BN) producing better quality of restoration images (RI1,..,RIK,..,RIN); wherein receptive field plays an important role in analyzing local features over degraded scenes. Another key feature of the proposed invention is the clipping of pre-defined multi-scale anchor boxes per cell to a restorated de-convolutional feature map (DC) only at the top of the network, which allows to efficiently reduce time-consumption. In terms of detection accuracy (recall-precision graph and mAP), the results of the reference dataset demonstrates the optimal performance of the proposed model and reveals the performance accuracy in low-light or rainy conditions to be higher than that in dusty or foggy conditions.

Type: Application

Filed: January 20, 2021

Publication Date: July 22, 2022

Applicants: Anu Singha (Ph.D Scholar, Department of Computer Science and Engineering, Tripura University), Sourav Dey Roy (Ph.D Scholar, Department of Computer Science and Engineering, Tripura University), Dr. Mrinal Kanti Bhowmik (Assistant Professor, Department of Computer Science & Engineering, Tripura University)

System and Method for Classification of Multiclass Scenes in Adversarial Weather using a Convolution Neural Network

New Indian Patent Application No: 202231068251

Status: Filed

Abstract: The present invention discloses a system and method for classification of multiclass scenes in adversarial weather using a convolution neural network. A CNN architecture is adopted, namely, AWDMC-Net for classifying scenes more accurately in adverse weather-degraded realistic scenes. The present invention relates to a one or more convolutional layer (Conv1…Convn) comprising a plurality of convolutional block (CB1, CB2, …. CBN). Another key feature of the proposed invention is adopting different combinations of skip connections in building blocks of CNN adaptively pruning the least convolutional kernels (3). “Entropy Guided Mean-l1 Norm” adaptively evaluate the convolutional kernels using the filters and their corresponding output feature maps (FMn). The prediction performance that represents six atmospheric/ weather conditions, namely, fog, dust, rain, haze, poor illumination, and clear day conditions. The AWDMC-Net reduces the time consumption for atmospheric/ weather classification tasks and meets the requirements of practical applications in real-world scenarios.

Type: Application

Filed: November 28, 2022

Applicants: Sourav Dey Roy (Ph.D Scholar, Department of Computer Science and Engineering, Tripura University), Anu Singha (Ph.D Scholar, Department of Computer Science and Engineering, Tripura University), Priya Saha (Ph.D Scholar, Department of Computer Science and Engineering, Tripura University) & Dr. Mrinal Kanti Bhowmik (Assistant Professor, Department of Computer Science & Engineering, Tripura University)

A System and a Method using GSNET for Presence of Gun in Complex Scenes

New Indian Patent Application No: 202231068259

Status: Filed

Abstract: A system and a method using Geographical and Semantic spatial-temporal Network (GSNet) for detecting presence of gun in complex scenes. The Geographical and Semantic spatial-temporal Network (GSNet) comprises a plurality of dense blocks to receive the at least one input image and connected to an at least one of small attention network. Each one of the small attention networks is used for both spatial attention and channel attention. An at least one of transition layer having one convolution layer to reduce the size of output feature map and the convolution layer imposes weight to a local patch which are relevant to the scene. The output feature map is inputted to an at least one of enhancement classification layer and output (R) to a softmax classifier to provide two output which displays the presence of gun.

Type: Application

Filed: November 28, 2022

Applicants: Rajib Debnath (Ph.D Scholar, Department of Computer Science and Engineering, Tripura University), Kakali Das (Ph.D Scholar, Department of Computer Science and Engineering, Tripura University), Priya Saha (Ph.D Scholar, Department of Computer Science and Engineering, Tripura University) & Dr. Mrinal Kanti Bhowmik (Assistant Professor, Department of Computer Science & Engineering, Tripura University)