1. Ph.D Theses

Permanent URI for this collectionhttps://idr.nitk.ac.in/handle/1/11

Browse

Filters

2018 - 201912020 - 20232

Settings

Subject: search.filters.subject.Deep Learning

Search Results

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    Plant Disease Detection Using Deep Learning-Based Approach
    (National Institute Of Technology Karnataka Surathkal, 2023) C K, Sunil; C D, Jaidhar; Patil, Nagamma
    Food security is threatening due to the exponentially growing global population. There are many reasons for food scarcity, such as exponential population, environmental dis- asters, climate change, the impact of COVID-19, and wars. Agriculture’s productivity has decreased in the last decade due to climate change and inappropriate usage of wa- ter, fertilizer, and pesticides, which stimulate plant diseases. Plant diseases and pests are also the cause of reducing the production of food all over the globe. Plant diseases cause around 20% to 40% loss in the production of agricultural products. Plant diseases extensively impact agrarian production growth. It results in a price hike on food grains and vegetables. Early detection of plant disease is essential to reduce economic loss and predict yield loss. Early perception of pathogens and insinuating proper medications are crucial to enhance crop yield and quality. Current plant disease detection involves the physical presence of domain experts to ascertain the disease. As a result, timely plant disease recognition entails sustained crop supervision from the start. Some research works have contemporarily been proposed as curative control measures. However, such an approach requires expensive equipment that is out of reach for small or middle-scale yeoman. Deep learning-based approaches vary in network architecture, and learning of the features by each model varies from one another in some aspects. To take this as an ad- vantage, this study proposed an ensemble-based deep learning approach using AlexNet, ResNet, and VGGNet. Seven different plant disease dataset is used with the binary and multiclass dataset. This ensemble-based approach enhances the classification result by minimizing the miss-classification effect. It constructively perceives plant diseases by analyzing plant leaf images. A broad set of experiments were conducted using differ- ent plant leaf image datasets such as Cardamom, Cherry, Grape, Maize, Pepper, Potato, and Strawberry to assess the agility of the proposed approach. Experiential results show that the proposed method attained a maximal detection accuracy of 100% for binary and 99.53% for multiclass datasets. Deep learning-based plant disease detection is proposed in this work by address- ing some of the challenges. Precise plant disease detection is essential, where more than one disease has similar symptoms and nature, and also to achieve excellent per- formance in spite of the imbalanced data. This study proposed a Multilevel Feature Fusion Network (MFFN), which combines the features learned at different levels of the network and also uses the adaptive attention technique by employing channel and pixel attention mechanism, which fabricates the network more robust by considering the ideeper network features which are shown in different channels and also with the pixel level features, with this the network is able to classify the diseases precisely on tomato plant dataset. The proposed deep learning-based approach is trained and tested on a tomato plant leaves dataset and achieved 99.36% training accuracy, 99.88% validation accuracy, and 99.5% external testing accuracy. It outperformed the existing approaches relevant to the tomato plant dataset. Further, this work also proposes a pesticide pre- scription module that provides pesticide information based on the type of tomato leaf disease. Plant disease detection using a complex background and images captured in differ- ent conditions is one of the challenges; this study proposed a cardamom plant disease detection approach by collecting the images in a complex background using different electronic gadgets. This study proposed a hybrid deep learning-based approach consist- ing of two stages: the background removal stage and the classification stage. U2 -Net is used for the background removal task, and EfficientNetV2 is used for the classifica- tion task. This makes the network more robust to handle the plant leaf images captured in complex nature.A large number of experiments were conducted to evaluate the pro- posed approach’s performance and compare it to other models such as EfficientNet and Convolutional Neural Network (CNN). According to the experimental results, the pro- posed approach achieved a detection accuracy of 98.26%. The approaches proposed in this study are producing prominent results. This study also suggested a pesticide prescription module for tomato plant leaf diseases. The pro- posed solutions in this study contribute to the field of plant disease detection, which can be adopted for real-time plant disease application. The overall aim of this study is to provide an efficient and robust plant disease detection approach.
  • Thumbnail Image
    Item
    Development Of Limited Supervised Deep Learning Methods For Biomedical Image Analysis
    (National Institute Of Technology Karnataka Surathkal, 2023) S J, Pawan; Rajan, Jeny
    Over the past few years, the computer vision domain has evolved and made a revo- lutionary transition from human-engineered features to automated features to address challenging tasks. Computer vision is an ever-evolving domain, having its roots deeply correlated with neuroscience. Any new findings that trigger a more intuitive under- standing and working of the human brain generally impact the design of computer vision algorithms. The convolutional neural network is one such algorithm that has become the de facto standard for most computer vision tasks, such as image classifi- cation, object detection, image segmentation, etc. However, the performance of CNNs is highly dependent on labeled data, making their practicability difficult in scenarios lacking sufficient labeled data, especially in medical applications. Therefore, it is im- perative to develop deep learning methods with limited supervision. In light of this, we explore the dimensions of deep learning with limited supervision through capsule networks and semi-supervised learning for biomedical image analysis, with a primary focus on segmentation. In this thesis, we have systematically reviewed various techniques for handling deep learning with limited labeled data, focusing on capsule networks and consis- tency regularization-driven semi-supervised learning. Capsule networks have shown immense potential for image classification tasks. However, extending it to pixel-level classification or segmentation is difficult. It poses numerous challenges, including the exponential growth of trainable parameters, expensive computation, and extensive memory overhead. In this regard, we propose DRIP-Caps, a Dilated Residual Incep- tion and Capsule Pooling framework that makes the capsule network lightweight by re- ducing the computation complexity without compromising performance on the CSCR (central Serous Chorioretinopathy) dataset. viiiSemi-supervised learning is a major discipline that alleviates the requirement for labeled data by incorporating labeled and unlabeled data to formulate pertinent infor- mation. We present a semi-supervised framework based on a mixup operation-driven consistency constraint for medical image segmentation by incorporating geometric con- straints regressing over the signed distance map (SDM) of the object of interest, achiev- ing superior performance on the publicly available ACDC and LA datasets. We also propose a novel semi-supervised framework for enforcing dual consistency (data level and network level) with the two-stage pre-training approach through networks of differ- ent learning paradigms enforcing both local and global semantic affinities, improving the overall performance. We envision these methods serving a major role in alleviating the tedious labeling process as far as the segmentation task is concerned.
  • Thumbnail Image
    Item
    Automatic Segmentation of Intra-Retinal Cysts from Optical Coherence Tomography Scans
    (National Institute of Technology Karnataka, Surathkal, 2018) G. N., Girish; Rajan, Jeny
    Retinal cysts are formed by accumulation of fluid in the retina caused by leakage due to blood retinal barrier breakdown from inflammation or vascular disorders. Analysis of retinal cystic spaces holds significance in detection, treatment and prognostication of several ocular diseases like age-related macular degeneration, diabetic macular edema, etc. Segmentation of intra-retinal cysts (IRCs) and their quantification is important for retinal pathology and severity characterization. In recent years, automated segmentation of intra-retinal cysts from optical coherence tomography (OCT) B-scans has gained significance in the field of retinal image analysis. In this thesis, a benchmark study is conducted to compare existing methods to identify the factors affecting IRC segmentation from OCT scans. A modular approach is employed to standardize the different IRC segmentation algorithms followed by analysis of variations in automated cyst segmentation performances and method scalability across image acquisition systems are done by using publicly available cyst segmentation challenge dataset (OPTIMA cyst segmentation challenge). Such exhaustive analysis on the scalability of OCT cyst segmentation methods in terms of methodological and input data variations has not been done before. An efficient cyst segmentation technique must be capable of performing cyst identification and delineation with minimum errors. Several methods proposed in the literature fail to delineate cysts up to their true boundary. To address this problem, an unsupervised vendor dependent method using marker controlled watershed transformation is proposed in this thesis. The method is based on two stages- k-means clustering technique is used to identify cysts in the form of marker, followed by topographical based watershed transform for final segmentation. Qualitative and quantitative evaluation of vthe proposed method is carried out against ground truth obtained from two graders on OPTIMA cyst segmentation challenge dataset (Spectralis Vendor OCT scans). Obtained results show that the proposed method outperformed other considered unsupervised methods. Several segmentation methods have been proposed in the literature for IRC segmentation on vendor-specific OCT images, but these lack generalizability across imaging systems. To address this issue, a fully convolutional network (FCN) model for vendor-independent IRC segmentation is proposed in this thesis. The proposed FCN was trained using the OPTIMA cyst segmentation challenge dataset (with four different vendor-specific images, namely, Cirrus, Nidek, Spectralis and Topcon). This method counteracts image noise variabilities and model over-fitting by data augmentation and hyper-parameter optimization. Additionally, sensitivity analysis of the model hyperparameters (depth and receptive field size) is performed to optimize the proposed FCN. The Dice Correlation Coefficient of the proposed method outperforms the algorithms published in the OPTIMA cyst segmentation challenge. Deeper FCNs exhibit better feature learning capabilities than shallower networks but those are computationally intensive due to large number of computation parameters and may be prone to vanishing gradient problem. To address this issue, a depthwise separable convolutional filter based end-end convolutional neural network architecture with swish activation functions is proposed in this thesis. OPTIMA cyst segmentation challenge dataset with four different vendor scans were used to evaluate the proposed architecture for vendor independent IRC segmentation task. Obtained experimental results show that the proposed method significantly reduced the number of computation parameters compared to regular convolution based FCN.