@sggs.ac.in
Assistant Professor
SGGS Institute of Engineering and Technology
Image Processing, Machine Learning, Data Mining
Scopus Publications
Scholar Citations
Scholar h-index
Scholar i10-index
Rimsha Taskeen Siddi Habib Hyder, Saba Siddiqua Sadiq Ahmed Siddiqui, Megha Jonnalagedda, and Arati Manjaramkar
Springer Nature Singapore
Saba Siddiqua Sadique Ahemed Siddiqui, Rimsha Taskeen Siddi Habib Hyder, Arati Manjaramkar, and Megha Jonnalagedda
IEEE
Lung diseases are becoming common everywhere in the world. These contain diseases like pneumonia, emphysema, tuberculosis, etc. which come under the category of Chronic Obstructive Pulmonary Diseases (COPDs). Other than these lung related diseases, a very popular recent lung disease is COVID-19. This work proposes an approach for detecting and classifying various lung related diseases with chest X-ray images using transfer learning concept, with MobileNet architecture as a feature extractor followed by Support Vector Machine (SVM). The images of chest X-ray are used to classify healthy individuals from patients suffering with various lung diseases like COVID-19, pneumonia, tuberculosis, and emphysema. In this study, a dataset of 2500 chest X-ray images of COVID-19, pneumonia, emphysema, tuberculosis infected, and normal images are used. The proposed method was able to achieve high accuracy 98.5% and F1-score of 99.8%.
Surekha Chalnewad and Arati Manjaramkar
IEEE
A license plate is alphanumeric rectangular plate. It is fixed on the vehicle and used for identification of the vehicle. Generally, huge numbers of vehicles move-on the road which is the major issue of concern in identifying the vehicle(s) owner, registration place of vehicle, address, etc. The automatic license plate detection is one of the solutions for such kind of problems. There are numerous methodologies available for license plate detection, but certain factors like speed of vehicles, language used on license plate, non-uniform letter effects on license plate, etc. makes the task of recognition difficult. The license plate detection system has many applications like payment of parking fees; toll fee on the highway; traffic monitoring system; border security system; signal system, etc. This research work proposes a novel license plate detection technique with the extension of Sobel mask. In proposed system, first step is acquisition of image. Second step is to detect the vehicle from the acquired image. In third step, segmentation of license plate from vehicle image is done. Finally, neural network classifier is used to classify the vehicle(s) license plate. The proposed system gives promising, robust, and efficient results for license plate detection. Proposed system achieves accuracy of 98% is achieved in detecting the license plate.
Shradha Itkare and Arati Manjaramkar
Springer Singapore
Arati Manjaramkar and Manesh Kokare
Springer Singapore
Incidences of diabetes are increasing worldwide. An eye complication associated with uncontrolled diabetes is diabetic retinopathy. If not treated, diabetic retinopathy can be vision threatening. The microaneurysms and dot hemorrhages are the only prominent clinically observable symptoms of DR. Their timely detection can help ophthalmologists in treating abnormalities efficiently and limit the disease severity. So detecting red lesions in early stage has become an indispensable task today. This paper gives an overview of earlier proposed algorithms and methods. It also compares these algorithms based on their performance for supporting the researchers by providing the gist of these algorithms. The standard retinal image databases are also compared and discussed.
Shaikh Rakhshinda Nahid M. Ayyub and Aarti Manjramkar
IEEE
Fruit Industry is the largest industry of India. Due to lack of maintenance, inappropriate manual inspection the fruit Disease causes huge losses in yield, quality and quantity. Manual inspection is tedious and time consuming process. An image processing approach is proposed for apple fruit disease identification and categorization using different color, texture and shape feature combination. The basic steps of the proposed approach are image segmentation, extraction of features (color, texture and shape), feature combination and finally apple disease identified and classified using multi-class support vector machine into diseased or normal class. Our proposed technique experimentally verified and validated. The accuracy of the proposed approach is achieved up to 96%.
Aashlesha Aswar and Arati Manjaramkar
Springer International Publishing
Abdul Rahman and Arati Manjaramkar
IEEE
Nowadays simultaneous extraction algorithms for finding common element sets does not allows concurrent execution, load steadiness, data allocation, and recovery mechanism on huge clusters. Therefore, we propose a Simultaneous Common Elements Extraction algorithm (SCEE) using MapReduce. To accomplish reduce storage and to prevent constructing traditional pattern bases, SCEE algorithm integrates the frequent items ultrametric tree, instead of traditional FP-trees. In this algorithm, the number of MapReduce tasks are three, designed to finish the whole extracting job. The mappers of third MapReduce job individually perform the decomposition operation on element sets, while the reducer will complete the grouping process by building ultrametric trees, and extraction of these trees distinctly. We also implemented an additional MapReduce job to find the top-K items from these frequent itemsets, we called this MapReduce job as an Aggregate Function. We designed the algorithm on cluster of computers. We demonstrate that the SCEE algorithm on the cluster is subtle to data allocation and dimension, since element sets with dissimilar length have different decomposition and building expenses. The real-world data for extensive experimentations prove that our proposed technique is effective and extendable.
Arati Manjaramkar and Manesh Kokare
Springer Science and Business Media LLC
Automated microaneurysm (MA) detection is still an open challenge due to its small size and similarity with blood vessels. In this paper, we present a novel method which is simple, efficient, and real-time for segmenting and detecting MA in color fundus images (CFI). To do this, a novel set of features based on statistics of geometrical properties of connected regions, that can easily discriminate lesion and non-lesion pixels are used. For large-scale evaluation proposed method is validated on DIARETDB1, ROC, STARE, and MESSIDOR dataset. It proves robust with respect to different image characteristics and camera settings. The best performance was achieved on per-image evaluation on DIARETDB1 dataset with sensitivity of 88.09 at 92.65% specificity which is quite encouraging for clinical use.
Arati Manjaramkar, , Manesh Kokare, and
The Intelligent Networks and Systems Society
Damage of retina due to diabetes is termed diabetic retinopathy. Hemorrhages and Microaneurysms are the first clinically visible symptoms of diabetic retinopathy. Detecting and treating diabetic retinopathy early can prevent vision loss. Accurate segmentation of retinal hemorrhage in color fundus image (CFI) has become a challenging task today; as retinal hemorrhages have varied size, shape and texture. We propose a connected component clustering method based on maximally stable extremal regions (MSER) for detecting many occurrences of hemorrhages with different shape and size in a fundus image. Proposed method has three main steps: firstly hemorrhage candidate generation, second is feature extraction and finally third step is hemorrhage detection. We have is evaluated our method on the DIARETDB1 and MESSIDOR dataset and experimental results show that the proposed system outperforms other state-of-the-art methods in detecting large and vessel connected hemorrhages. The proposed method achieves image level sensitivity, specificity of 96.45, 97.64 and lesion level sensitivity, specificity of 94.89, 98.9 respectively.
Arati Manjaramkar and Manesh Kokare
IEEE
Detection of diabetic retinopathy (DR) in its early stage has become a crucial task among researchers because symptoms of DR can be perceived by patient only in severe stage. The purpose of this research is to detect DR in early stage. Early detection decrease blindness prevalence. Microaneurysm (MA) are the only earliest noticeable feature of DR. We propose a novel two-level method: coarse level use morphology for MA candidate detection and fine level use a binary decision tree (DT) classifier based on classification and regression tree(CART) for true MA classification. The proposed CART classifier discriminates MA and non-MA pixels efficiently. A 10-fold cross-validation using CART is done on standard public dataset DIARETDB1 for MA classification. The proposed system achieves a promising classification accuracy of 98.6%. Results demonstrate that proposed system performance is comparable to clinical experts.
Umesh S. Bhoskar and Arati Manjaramkar
IEEE
One of the essential tasks in data integration is entity resolution (ER) which will recognize the records that are belonging to the same entity. The entity resolution is referred by many other terms like duplicate detection, pattern matching, etc. Now a days the activities like information integration, information retrieval, crowd sourcing, and pay-as-you-go have involved users to carry out the ER tasks such as to identify whether two entity descriptions are referred to the same entity or not. Previous work of ER involves clustering and comparison approaches which are based on some assumption. The ER gives the poorer quality when such assumptions are not correct. In our approach, we present a new set of entity rules where each rule enumerates all possibilities to identify the correct entity of the records. Additionally, we propose an extended approach (GenR) for efficient and effective rules generation by using a specialized form of term-based entropy measure. We experimentally evaluated the proposed approach using data set with a large no. of records and the data sets with different data characteristics. We report on some promising empirical results which demonstrate performance improvement by using a term-based quality measure.
Pradeep R. Dumne and Arati Manjaramkar
IEEE
In MANET, node changes its location so it adopt property like infrastructure-less. A group of mobile nodes which are agree and capable to establish relations, using without any fixed centralize supervision and infrastructure of the network is known as mobile ad hoc network. To start communications among the mobile nodes, they should be available to each other. But presence of malicious nodes, might fail this requirement and interrupt integrity, confidentiality, availability of network services such as routing processes. So prevention and detection of malicious nodes is a major task. In this paper, we proposed a method to resolve this problem by using malicious node detection schema based upon DSR mechanism-cooperative bait detection scheme (CBDS) which uses hybrid defense architectures. CBDS technique helps to find out malicious node by using a reverse tracing technique. The basic and proposed CBDS schemes are implemented in NS-2.35. Results are examined on the basis of throughput, PDR.
Arati Manjaramkar and Rahul L Lokhande
IEEE
Many web databases contains the data in the form of structured, semi structured and in unstructured format. This paper studies the issue of extracting these data records from online web database. The main motto of this paper is to recognize the data region which contains the data records, divide these data records, mine the data value from them and keep these extracted record in a structured format. This arrangement of extracted data is useful for many application like knowledge discovery purpose etc. Existing system has some data records arrangement problem which does not arrange dynamically generated web data properly. The proposed system is based on identification of data records, extraction of data values and arranging these data values in a database. The proposed system uses the partial tree alignment method for giving the better alignment outcome.
Arati Manjaramkar and Manesh Kokare
IEEE
Inefficiency of human body to produce and consume insulin leads to diabetes. This diabetes over a period of time starts showing adverse effects on different organs. If it affects eye it is termed as diabetic retinopathy (DR). Earlier researchers have put forth many automated microaneurysms detection systems, but none of them can replace clinical procedure. Here, we propose an rule based system for microaneurysms detection in digital fundus images. The proposed system is a three step algorithm. Initially image pre-processing is done, secondly candidate microaneurysms are segmented, thirdly features are extracted from these candidates and true microaneurysms are recognized using rule based expert system. The system performance is evaluated on publicly available database DIARETDB1 which consists of evaluation protocol and ground truth collected from experts. Overall Sensitivity of 80.6% at 97.50% specificity with accuracy of 95.95% is achieved by proposed method.