Nirmala G

@mahendra institute of technology

Associate Professor /Department of Electronics and Communication Engineering
Mahendra Institute of Technology

Nirmala G


                  Download Compact PDF  
https://researchid.co/nirmalaguru

RESEARCH, TEACHING, or OTHER INTERESTS

Computer Vision and Pattern Recognition, Artificial Intelligence, Cancer Research, Hardware and Architecture
11

Scopus Publications

Scopus Publications

  • Automatic cervical cancer classification using adaptive vision transformer encoder with CNN for medical application
    G. Nirmala, P. Prathap Nayudu, A. Ranjith Kumar, Renuka Sagar
    Pattern Recognition, 2025
  • Polycystic ovary syndrome detection using optimized SVM and DenseNet
    E. Silambarasan, G. Nirmala, Ishani Mishra
    International Journal of Information Technology Singapore, 2025
  • ADRAN: Adaptive Deep Residual Aggregation Network-based Cervical Region Segmentation in Cervix Images
    G. Nirmala, P. Suresh Kumar, Anand Karuppannan
    IETE Journal of Research, 2025
  • Real-Time Sign Language Translator with IoT and Flex Sensors for Multilingual Support
    Sudhan.M.B, Senthil Kumar K, Usha Bala Varanasi, G. Nirmala, Abijith G R, G. Karthikeyan
    Proceedings of 7th International Conference on Inventive Material Science and Applications Icima 2025, 2025
    Sign language functions as a fundamental communication system for deaf individuals while population-wide ignorance about sign language presents a major obstacle. Sign language translation systems available today deal with performance problems that include long delays between communication, restricted language availability, and poor recognition of natural hand movements. The proposed real-time IoT-based sign language translator equips a smart wearable flex and IMU sensor-embedded glove for sign language translation purposes. A modified LSTM model operates within the system for gesture recognition and delivers 97.2% precision through a 180ms translation response period. The cloud system supports extended language options through its 10+ multilingual speech features. The user experience evaluation met a rating of 4.8 out of 5 while the device operated for 12 hours on a single charge. The proposed system enhances both latency speed by 30% and classification precision compared to earlier models. This invention delivers a customizable system that enables instant accessible communication.
  • Deep Learning-Augmented Hybrid Optimization for Crop Classification using Hyperspectral images
    J. Mohana, G. Nirmala, Praveen Damacharla, Mohammad Bdair, R. Logarasu, S. Karpagam
    International Conference on Distributed Systems Computer Networks and Cybersecurity Icdscnc 2024, 2024
    Precise and efficient solutions are necessary for the essential job of classifying crops using hyperspectral images in precision agriculture. The paper introduces a framework called Deep Learning-Augmented Hybrid Optimization (DLAHO), which combines a CNN for feature extraction with a hybrid optimization technique that combines Genetic Algorithms (GA) and Gradient-Based Optimization (GBO) to optimize model parameters. The Hyperspectral Image Database (HID) preprocesses hyperspectral images by using normalization and PCA to reduce dimensionality. The CNN utilizes both spectral and spatial data, which are then enhanced by the hybrid GA-GBO method. The improved model accurately identifies crops with a high level of precision. The results indicate that DLAHO obtains an accuracy of 95.4% for wheat, 92.1% for maize, 94.3% for soybean, 93.2% for barley, 96.0% for rice, 91.5% for sugarcane, 93.8% for potato, and 92.7% for cotton. DLAHO outperforms baseline models and state-of-the-art algorithms, achieving an F1-score of 93.1%, recall of 93.5%, precision of 92.8%, and total accuracy of 93.6%. The findings show that the recommended DLAHO architecture significantly increases the precision and dependability of crop categorization, thereby making it a viable tool for precision agriculture.
  • Retraction Note to: A novel bat optimized runlength networks (BORN) for an efficient classification of breast cancer (Journal of Ambient Intelligence and Humanized Computing, (2021), 12, 5, (4797-4808), 10.1007/s12652-020-01890-7)
    G. Nirmala, P. Suresh Kumar
    Journal of Ambient Intelligence and Humanized Computing, 2023
  • Realization of Human Eye Pupil Detection System using Canny Edge Detector and Circular Hough Transform Technique
    Srikrishna M, Nirmala G
    Proceedings of the 2nd International Conference on Applied Artificial Intelligence and Computing Icaaic 2023, 2023
    Near Infrared (NIR) images involves the generation of an edge-map by combining two edge-maps generated from the same eye image for pupil detection. It is accomplished by the use of Gaussian filtering, picture binarization, and Sobel edge detection techniques. Image segmentation is used to group similar pixels based on the rate of change in intensity or depth, allowing for the representation of information from the image. The Hough transformation is employed as an efficient method for detecting lines in images, with this work proposing the use of angle-radius parameters instead of slope-intercept parameters, simplifying computation and facilitating pupil detection. This approach increases the accuracy and speed of pupil recognition by reducing erroneous edges in the edge-map. This technique's hardware implementation on an FPGA platform may be utilized for recognition and iris localization applications.
  • Performance Enhancement of Cognitive Radio Networks Using SINR-Based Cooperative Beamforming
    P. Suresh Kumar, G. Nirmala, S. Manimegalai, H. Arulvedi
    International Journal of Advanced Science and Engineering, 2022
    The Model Cognitive Radio Network (MCRN) is developed as a cooperative beam forming method specially designed to provide dynamic spectrum allocation for Primary and Secondary Users. The purpose of the design is to minimize the transmit power of a transmitter area, an eavesdropper, a primary receiver, and multiple secondary receivers were given with distinct Signal to Interference and Noise Ratio (SINR). This method easily detects noises without any prior knowledge about the signal. The signal fading is avoided using the technique was relay between the users, this leads to result in the reduction of outage probability. The performance estimation of the proposed optimal approach is validated and compared using the computer simulations software with an accuracy of 99.8% of the existing methods.
  • Evaluation of Topography Retrieval of Exo Mars Using Single Image DTM Estimation and Super Resolution Restoration Method-Review
    P. Suresh Kumar, G. Nirmala, S. Manimegalai, H. Arulvedi
    International Journal of Advanced Science and Engineering, 2022
    This review examines the meta-data of high-resolution orbital imagery obtained by Mars during the last four decades. The aim of this study is to provide a starting point for planetary scientists interested in exploring the martian surface for modifications linked to natural phenomena. A framework for generating picture groupings relevant to prioritising regions for shift detection is adopted and used for analysis. The season, the Martian Year, and the local period that an image was captured, as well as the imaging device and its resolution, are the criteria that determine each grouping. The present work indicates that there is enough coverage to regularly analyze periodic martian phenomena in images depicting the same region during the same season, as well as intermittent martian phenomena in images depicting the same area in various time periods. While with this treatment of the human visual system is short, it offers a simple understanding of the eye's capabilities in perceiving pictorial details. To clean up the noise and other defects in the data, to improve subtle information not noticeable in batch-processed NASA photos were discussed. To magnify smaller features for study, the used fairly standard image processing techniques.
  • A novel bat optimized runlength networks (BORN) for an efficient classification of breast cancer
    G. Nirmala, P. Suresh Kumar
    Journal of Ambient Intelligence and Humanized Computing, 2021
  • An efficient detection of structural similarity in mammograms using support vector machine (SVM) classifier
    International Journal of Scientific and Technology Research, 2020