Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Maylor K. H. LeungSuggested from co-authorship
- Michael S. BrownSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works186 from public data
- Video Processing From Electro-Optical Sensors for Object Detection and Tracking in a Maritime Environment: A Survey493
A survey on maritime object detection and tracking approaches, which are essential for the development of a navigational system for autonomous ships, is presented, followed by an approach-based taxonomy wherein the advantages and limitations of each approach are compared.
- Illuminant estimation for color constancy: why spatial-domain methods work and the role of the color distribution367
It is shown that spatial information does not provide any additional information that cannot be obtained directly from the color distribution and that the indirect aim of spatial-domain methods is to obtain large color differences for estimating the illumination direction.
- Training-Based Spectral Reconstruction from a Single RGB Image222
A training-based method to reconstruct a scene’s spectral reflectance from a single RGB image captured by a camera with known spectral response, based on a radial basis function network that leverages RGB white-balancing to normalize the scene illumination to recover the scene reflectance.
- ElliFit: An unconstrained, non-iterative, least squares based geometric Ellipse Fitting method170
The results show that the proposed ellipse fitting method is quite selective to elliptic shapes only and provides accurate fitting results, indicating potential application in medical, robotics, object detection, and other image processing industrial applications.
- Neural Network Based Country Wise Risk Prediction of COVID-19163
A shallow long short-term memory (LSTM) based neural network to predict the risk category of a country and shows that the proposed pipeline outperforms state-of-the-art methods for data of 180 countries and can be a useful tool for such risk categorization.
- Edge curvature and convexity based ellipse detection method145
The proposed ellipse detection method performs far better than existing methods and is close to the ideal results, with precision, recall, and F-measure, all very close to 1.
- A novel framework for making dominant point detection methods non-parametric94
A theoretical bound of the maximum deviation of pixels obtained by digitization of a line segment is used for constructing a general framework to make most dominant point detection methods non-parametric.
- Object Detection in a Maritime Environment: Performance Evaluation of Background Subtraction Methods81
- Prediction of porosity and thermal diffusivity in a porous fin using differential evolution algorithm71
Two numerical examples provide engineering insight into the problem of designing porous fins using good thermal conductors like aluminum and copper along with the working of DE by using the differential evolution (DE)-based optimization technique.
- Physics-based machine learning for subcellular segmentation in living cells68
A physics-based simulation approach is used to train neural networks to automatically segment subcellular structures despite the optical artefacts, and extensive results on the segmentation of small vesicles and mitochondria in diverse and independent living- and fixed-cell datasets are shown.
- High spatially sensitive quantitative phase imaging assisted with deep neural network for classification of human spermatozoa under stressed condition60
A partially spatially coherent digital holographic microscope (PSC-DHM) for quantitative phase imaging (QPI) is developed in order to distinguish normal sperm cells from sperm cells under different stress conditions such as cryopreservation, exposure to hydrogen peroxide and ethanol.
- 57
- 56
- Accurate detection of ellipses with false detection control at video rates using a gradient analysis54
A novel fast and robust ellipse detection method that adopts arcs selection, smart grouping, and repeated utilization of gradient information to significantly reduce the computations otherwise needed without compromising the detection effectiveness.
- Physics-Guided Loss Functions Improve Deep Learning Performance in Inverse Scattering53
New designs of loss functions are proposed which incorporate multiple-scattering based near-field quantities (such as scattered fields or induced currents within domain of interest) to solve electromagnetic inverse scattering problems.
- Single Image Dehazing for a Variety of Haze Scenarios Using Back Projected Pyramid Network51
A novel generative adversarial network architecture for this problem, namely back projected pyramid network (BPPNet), that gives good performance for a variety of challenging haze conditions, including dense haze and inhomogeneous haze.
- A COMPUTATIONAL MODEL OF MACHINE CONSCIOUSNESS51
A physical definition of consciousness is proposed with the hope to model it in intelligent machines and a concept of mental saccades that is useful for explaining the attention-switching and focusing mechanism from computational perspective is proposed.
- 45
- 44
- MuSCoWERT: multi-scale consistence of weighted edge Radon transform for horizon detection in maritime images43
This paper presents a novel method called multi-scale consistence of weighted edge Radon transform, abbreviated as MuSCoWERT, which detects the long linear features consistent over multiple scales using multi- scale median filtering of the image followed byRadon transform on a weighted edge map and computing the histogram of the detected linear features.
- 43
- Client Selection in Federated Learning under Imperfections in Environment40
A novel sampling method named the irrelevance sampling technique that selects a subset of clients based on quality and quantity of data on edge devices that achieves 50–80% faster convergence even in highly skewed data distribution in the presence of free riders.
- SoCodeCNN: Program Source Code for Visual CNN Classification Using Computer Vision Methodology40
This research is the first to propose a novel human-inspired approach to automatically convert program source-codes to visual images that could be then utilized for automated classification by visual convolutional neural network (CNN) based algorithm.
- 38
- 35
- Classification of Micro-Damage in Piezoelectric Ceramics Using Machine Learning of Ultrasound Signals34
The results show that appropriate feature design combined with simple k-nearest neighbor classifier can provide up to 98% classification accuracy even though conventional features for time-series data and a variety of classifiers cannot achieve close to 70% accuracy.
- Deep learning architecture “LightOCT” for diagnostic decision support using optical coherence tomography images of biological samples32
A custom deep learning architecture, LightOCT, is proposed for the classification of OCT images into diagnostically relevant classes and is shown to provide excellent training and test results for diverse OCT image datasets.
- Are Object Detection Assessment Criteria Ready for Maritime Computer Vision?32
New bottom edge proximity metrics are considered as assessment metrics for maritime computer vision and indicate that existing computer vision approaches are indeed promising for maritimeComputer vision and can play a foundational role in the emerging field of Maritime computer vision.
- Dense Video Captioning: A Survey of Techniques, Datasets and Evaluation Protocols31
All of the studies that claim to perform DVC along with its sub-tasks and summarize their results are discussed and all of the datasets that have been used for DVC are discussed.
- Emotionally charged text classification with deep learning and sentiment semantic31
This paper shows that documents can be represented as a sequence of vectors carrying semantic meaning and classified using a recurrent neural network that recognizes long-range relationships and shows that in this representation, additional sentiment vectors can be easily attached as a fully connected layer to the word vectors to further improve classification accuracy.
- MSCM-LiFe: Multi-scale cross modal linear feature for horizon detection in maritime images31
This paper demonstrates the performance of the proposed multi-scale cross modal linear feature for horizon detection over 13 videos and shows that the proposed method detects horizon with small error in most of the cases, outperforming three state-of-the-art methods.
- 27
- 27
- Sentiment analysis using EEG activities for suicidology26
This study provides an important and interesting conclusion that prior experience or conditioned training in detection of suicide notes is not beneficial in discriminating between genuine and hoax suicide notes and that the brain signals are more discriminatory in participants with no prior experience and conditioning.
- 26
- Enabling grasp action: Generalized quality evaluation of grasp stability via contact stiffness from contact mechanics insight24
A new evaluation approach of the quality of grasping stability via constructing a model of grasping stiffness based on the theory of contact mechanics is presented, validates the proposed approach and designs two grasping cases to verify the proposed measurement criterion.
- 23
- ELM-HTM guided bio-inspired unsupervised learning for anomalous trajectory classification23
A new bio-inspired learning model for a single-class classifier to detect abnormality in video object trajectories using a simple but dynamic extreme learning machine (ELM) and hierarchical temporal memory (HTM) together referred to as ELM-HTM in an unsupervised way to learn and classify time series patterns.
- Motivation detection using EEG signal analysis by residual-in-residual convolutional neural network22
The results clearly indicate the potential of detecting motivation states using EEG signals, provided suitable methodologies such as proposed in this paper, are employed.
- Label-free non-invasive classification of rice seeds using optical coherence tomography assisted with deep neural network21
A fast, non-contact and non-invasive technique, deep learning assisted optical coherence tomography (OCT) for subsurface imaging in order to distinguish different seed varieties, which can accurately classify seed varieties irrespective of the morphological similarities.
- High-resolution visualization and assessment of basal and OXPHOS-induced mitophagy in H9c2 cardiomyoblasts20
High-resolution imaging approaches applied on H9c2 cells provide novel insights on mitophagy during physiologically relevant conditions and demonstrate the importance of both canonical as well as non-canonical autophagy mediators in lysosomal degradation of mitochondria after OXPHOS induction.
- 20
- PIE-RSPOP: A brain-inspired pseudo-incremental ensemble rough set pseudo-outer product fuzzy neural network20
Despite the usually uninterpretable nature of incremental ensembles of fuzzy networks, a scheme to derive simple interpretable single knowledge base is also proposed to demonstrate the advantages of PIE-RSPOP over other incremental learning methods.
- Metrics and statistics of frequency of occurrence of metamerism in consumer cameras for natural scenes19
This paper presents metrics and statistics of frequency of occurrence of metamerism in three consumer cameras, viz., Canon 1D Mark III, Nikon D40, and Sony α7, using spectral and RGB images of natural scenes and uses angular difference and digital equivalence approaches for this purpose.
- 19
- Single-shot multispectral quantitative phase imaging of biological samples using deep learning18
The aim of the present study is to extract spectral dependent quantitative information in single-shot using a highly spatially sensitive digital holographic microscope assisted by a deep neural network to generate multispectral quantitative phase maps from a single input interferogram.
- IT2-GSETSK: An evolving interval Type-II TSK fuzzy neural system for online modeling of noisy data18
Interval Type-II version of GSETSK (Generic Self-Evolving Takagi-Sugeno-Kang), namely IT2-GSETSK is presented, which is a fully-online model able to handle data streams and cope with time-variant data and provides up-to-date, relevant and compact rule base that is easily interpretable by human.
- 18
- 17
- 17
- Virtual labeling of mitochondria in living cells using correlative imaging and physics-guided deep learning16
The results show that the virtual labeling approach is a powerful way of segmenting and tracking individual mitochondria in bright-field images, results previously achievable only for fluorescently labeled mitochondria.
- 16
- 15
- An Error Bounded Tangent Estimator for Digitized Elliptic Curves15
The ellipse detection method is greatly benefited by the proposed tangent estimator, as the maximum error in geometrical ellipSE detection is no more critically dependent upon the tangent estimation (due to the reduced error in tangent estimating).
- RS-HeRR: a rough set-based Hebbian rule reduction neuro-fuzzy system14
This paper combines a powerful rule generation and reduction fuzzy system, called Hebbian-based rule reduction algorithm (HeRR) and a novel rough-set-based attribute selection algorithm for rule reduction, named RS-HeRR, for the generation of effective, interpretable and compact rule set.
- 14
- 13
- Recent Advances in Statistical Data and Signal Analysis: Application to Real World Diagnostics from Medical and Biological Signals13
This special issue was an attempt to bring together interesting works that use advanced statistical techniques for cutting edge medical applications and biological signals for disease detection and diagnosis and received submissions from a wide range of approaches and applications.
- A Precise Ellipse Fitting Method for Noisy Data13
A least squares method that does not require a constrained optimization and has very low false positive rates is proposed that is superior performance than most least squares based method for elliptic curves and greater true negative rates for non-elliptic curves even in presence of 30% noise.
- 13
- Data-Efficient Training of CNNs and Transformers with Coresets: A Stability Perspective12
It is demonstrated that in the absence of the right pretraining, CNNs are better at learning the semantic coherence between spatially distant objects within an image, and these tend to outperform transformers at almost all choices of the coreset size.
- 12
- 12
- Assessing Error Bound For Dominant Point Detection12
The assessment presented in this paper will enable the users of dominant point detection methods to understand the nature of error in the method of their choice and help them to make better decision about the methods and their control parameters.
- A hybrid approach for ellipse detection in real images12
A new method of ellipse detection is proposed, which is able to deal with the challenges mentioned above, while being computationally efficient and more accurate than existing methods.
- High-resolution imaging in acoustic microscopy using deep learning11
Deep learning-based high-resolution imaging in acoustic microscopy is proposed in this paper and is relevant to a wide range of industrial applications, including electronic production, material micro-structure analysis, and other biological applications in general.
- BMWP: the first Bengali math word problems dataset for operation prediction and solving10
This pioneering study introduces the first Bengali Math Word Problem Benchmark Data Set (BMWP), comprising 8653 word problems, and investigates operation prediction from Bengali word problems using state-of-the-art deep learning (DL) techniques.
- Object detection neural network improves Fourier ptychography reconstruction10
This work derives how illumination estimation can be cast as an object detection problem that permits the use of a fast convolutional neural network (CNN) for this task and finds that faster-RCNN delivers highly robust results and outperforms classical approaches by far with an up to 3-fold reduction in estimation errors.
- 10
- Propagating Transparency: A Deep Dive into the Interpretability of Neural Networks9
This paper offers a comprehensive overview of interpretable deep learning methods, and emphasizes gradient-based propagation techniques that shed light on the complex mechanisms driving neural network predictions.
- 9
- Temporal Motionless Analysis of Video using CNN in MPSoC9
This is the first work on TMAV using Convolutional Neural Network for scene prediction in MPSoCs and a metric named, Energy Consumption per Training Image (ECTI) is introduced to assess the suitability of using a CNN model in mobile MP soCs with a focus on energy consumption of the device.
- 9
- Online tracking of deformable objects under occlusion using dominant points9
A novel nonintegral time propagation model for propagating the dominant points of the boundary contour using a linear operation and an analytical conjugate gradient approach for online robust learning of the shape deformation and motion model is proposed.
- 9
- A Perspective on Machine Consciousness9
Various perspectives related to cognitive and information science are considered to conclude that consciousness is a physical phenomenon.
- 8
- 8
- Counterfactual Explainable Gastrointestinal and Colonoscopy Image Segmentation8
An optimized deep learning solution using DeepLabv3+ with ResNet-101 as its backbone provides improved and reliable quantitative and qualitative results in comparison to other state-of-the-art (SOTA) methods on two publicly available gastrointestinal and colonoscopy datasets.
- IRON-MAN: An Approach to Perform Temporal Motionless Analysis of Video Using CNN in MPSoC8
This is the first work on Temporal Motionless Analysis of the Video using Convolutional Neural Network (CNN) for scene prediction in MPSoCs and experimental results show that the methodology outperforms state-of-the-art.
- 8
- Fast segmentation of sub-cellular organelles8
The proposed method uses a hybrid combination of several image processing techniques and is effective in segmenting the sub-cellular structures in a fast and effective manner.
- A Non-heuristic Dominant Point Detection Based on Suppression of Break Points8
This paper presents a non-heuristic and control parameter independent dominant point detection method based on the suppression of break points and provides a more balanced performance than the recent break point suppression methods.
- Deep Learning-Based Denoising of Acoustic Images Generated With Point Contact Method7
This paper bench-marked and compared several state-of-the-art deep learning image denoising methods with the classical methods on ultrasonic images, and demonstrated the effectiveness and versatility of the deep learning-based Denoising model for the unexplored domain of ultrasound/ultrasonic data.
- UltraMNIST Classification: A Benchmark to Train CNNs for Very Large Images7
A novel research problem of training CNN models for very large images, and a simple yet representative benchmark dataset for this task, which is intended to pave the ground for a new generation of CNN methods suitable for handling large images in an efficient and resource-light manner.
- Biosignal-Based Driving Skill Classification Using Machine Learning: A Case Study of Maritime Navigation7
The proposed algorithm showed that the biosingal data associated with the experts can be categorized as different from that of the novices, which is in line with the results of NASA Task Load Index (NASA-TLX) rating scores.
- Topic-based Video Analysis7
The scope, variations, and applications of topic modelling, particularly focusing on surveillance video analysis, are discussed, with a methodological survey on existing topic models, their features, underlying representations, characterization, andApplications in visual surveillance’s perspective.
- Fabrication imperfection analysis and statistics generation using precision and reliability optimization method7
It is shown that PRO can clearly highlight several image effects and imperfections which may not be easily identifiable by human eye or may be difficult to analyze and assess and can be used to generate useful fabrication accuracy statistics and detect features with low fidelity or more imperfections.
- PRO: A Novel Approach to Precision and Reliability Optimization Based Dominant Point Detection7
A novel method that uses both the local and the global nature of fit for dominant point detection is proposed, which is especially useful for segmentation, shape representation, activity recognition, and robust edge feature extraction in object detection and recognition problems.
- Patch Gradient Descent: Training Neural Networks on Very Large Images6
Patch Gradient Descent is based on the hypothesis that instead of performing gradient-based updates on an entire image at once, it should be possible to achieve a good solution by performing model updates on only small parts of the image at a time, ensuring that the majority of it is covered over the course of iterations.
- 6
- 6
- 6
- Classification of Hyperspectral or Trichromatic Measurements of Ocean Color Data into Spectral Classes6
A method for classifying radiometric oceanic color data measured by hyperspectral satellite sensors into known spectral classes, irrespective of the downwelling irradiance of the particular day, i.e., the illumination conditions is proposed.
- High Availability Based Migration Analysis to Cloud Computing for High Growth Businesses6
This study provides a paradigm and a useful framework for analyzing cloud computing services and presents an analysis of the important factors influencing the decision to migrate to networks supporting scalability and high availability.
- Multilingual visual question answering for visually impaired people5
Two VQA datasets in Bengali and Hindi are introduced, machine-translated from the popular VQA-VI dataset VizWiz, and curated by native speakers to create a baseline for VQA for VI people in resource-constrained Indic languages.
- Towards a More Inclusive AI: Progress and Perspectives in Large Language Model Training for the Sámi Language5
This is the first study on the Sami language for adapting non-statistical language models that use the latest developments in the field of natural language processing (NLP).
- An UltraMNIST classification benchmark to train CNNs for very large images5
The ‘UltraMNIST dataset’ and associated benchmarks for this new research problem of ‘training CNNs for large images’ are presented and two variants of the problem are discussed: standard version that facilitates the development of novel CNN methods for effective use of the best available GPU resources and the budget-aware version to promote the development of methods that work under constrained GPU memory.
- Auxiliary Network: Scalable and Agile Online Learning for Dynamic System with Inconsistently Available Inputs5
A novel deep learning-based model called Auxiliary Network (Aux-Net), which is scalable and agile, which employs a weighted ensemble of classifiers to give a final outcome.
- Latent Graph Attention for Enhanced Spatial Context5
This paper presents Latent Graph Attention (LGA), a computationally inexpensive and stable, modular framework for incorporating the global context in the existing architectures, especially empowering small-scale architectures to give performance closer to large size architectures, thus making the light-weight architectures more useful for edge devices with lower compute power and lower energy needs.
- CARF-Net: CNN attention and RNN fusion network for video-based person reidentification5
A shallow end-to-end network which incorporates two stream convolutional neural networks, discriminative visual attention and recurrent neural network with triplet and softmax loss to learn the spatiotemporal fusion features is proposed.
- Gamut expansion of consumer camera to the CIE XYZ color gamut using a specifically designed fourth sensor channel5
The proposed approach leverages on the matrix-R theory that states that the space of metamerism of a sensor, known as the metameric black space, can be determined directly from the camera's spectral sensitivities.
- 5
- Latent Graph Attention for Spatial Context in Light-Weight Networks: Multi-Domain Applications in Visual Perception Tasks4
Latent Graph Attention is presented, a computationally inexpensive (linear to the number of nodes) and stable modular framework for incorporating the global context in existing architectures and improves performance in three challenging applications, namely transparent object segmentation, image restoration for dehazing and optical flow estimation.
- Online Learning under Haphazard Input Conditions: A Comprehensive Review and Analysis4
This survey paper addresses the topic of online learning in the context of haphazard inputs in the context of haphazard inputs, explicitly foregoing such an assumption of the input feature space of the streaming data.
- Image inpainting in acoustic microscopy4
A deep learning-enabled image inpainting for acoustic microscopy is proposed based on training various generative adversarial networks to inpaint holes in the original image and generate a 4× image from it.
- Analyzing Mitochondrial Morphology Through Simulation Supervised Learning4
A machine learning-aided segmentation and analysis pipeline is demonstrated for the quantification of mitochondrial morphology in fluorescence microscopy images of fixed cells and eliminates the requirement for ground truth annotations for supervised deep learning.
- Can we automate diagrammatic reasoning?4
A diagrammatic reasoning dataset is presented that provides a large variety of DR problems and a Knowledge-based Long Short Term Memory (KLSTM) is proposed to solve diagrammatic Reasoning problems.
- Artefact removal in ground truth deficient fluctuations-based nanoscopy images using deep learning4
This work proposes a robust and versatile simulation-supervised training approach of deep learning auto-encoder architectures for the highly challenging nanoscopy images of sub-cellular structures inside biological samples and investigates the scope of generalizability across structures, andnanoscopy algorithms not included during simulation- supervised training.
- 4
- 4
- 4
- VisionAidQA: Advancing Visual Question Answering for the Visually Impaired3
An in-depth review of the state-of-the-art VQA systems, datasets, and methodologies, focusing on their application to assist VI users, to provide valuable insights and guidance for researchers and developers working on VQA technologies for VI people.
- 3
- RCV2023 Challenges: Benchmarking Model Training and Inference for Resource-Constrained Deep Learning3
- Digital Staining of Mitochondria in Label-free Live-cell Microscopy3
A new loss function with multiple thresholding steps is proposed to promote more effec-tive learning for microscopy data and demonstrate a deep learning approach to translate the label-free brightfield images of living cells into equivalent fluorescence images of mitochondria with an average structural similarity of 0.77.
- T-MIS: Transparency Adaptation in Medical Image Segmentation3
This work proposes image-specific fine-tuning to make a deep learning model adaptive to specific medical imaging tasks and shows that the proposed model is more robust to segment previously unseen objects (negative test dataset) than state-of-the-art CNNs.
- Polygonal Approximation of Digital Planar Curve Using Novel Significant Measure3
The proposed method differentiates between the situations when a point on the curve between two points on a curve projects directly upon the line segment or beyond this line segment, and allows preservation of points with high curvature even as revised set of dominant points are derived.
- 3
- Non-heuristic automatic techniques for overcoming low signal-to-noise-ratio bias of localization microscopy and multiple signal classification algorithm3
Techniques to automatically debias localization microscopy and multiple signal classification algorithm of these biases without compromising their resolution and without employing heuristics, user-defined criteria are presented.
- 3
- Design of macro-filter-lens with simultaneous chromatic and geometric aberration correction3
A macro-filter-lens design that can correct for chromatic and geometric aberrations simultaneously while providing for a long focal length is presented and provides an inexpensive and optically good (aberration compensated) solution for macro imaging of objects not placed close to the camera.
- IReCAPTCHA: a robust image-reasoning CAPTCHA system2
This work proposes a novel image-reasoning CAPTCHA, called IReCAPTCHA, which integrates image understanding, noise mitigation, and mathematical reasoning, and suggests that the approach has strong potential to enhance online security by effectively deterring automated access attempts by malicious bots.
- Enhanced Image Steganography Using Secure Random Pixel Distribution2
SRPD is a steganographic system that is secure, imperceptible, and cost‐effective in terms of computational resources; thus, it can be used for secret communication and privacy‐preserving multimedia applications.
- 3D Mitochondria Shape Library for Optical Microscopy (3DMSL): A multimodal dataset for deep learning based mitochondrial analysis2
3DMSL is released, a database of 3D shapes of mitochondria, a database of diverse mitochondria shapes in different 3D shape representation formats of meshes, point clouds and implicit shapes, for microscope analysis of cell organelles.
- Explain with Confidence: Fusing Saliency Maps for Faithful and Interpretable Weakly-Supervised Model2
- Hedging Is Not All You Need: A Simple Baseline for Online Learning Under Haphazard Inputs2
This work proposes HapNet, a simple baseline that is scalable, does not require online backpropagation, and is adaptable to varying input types, and introduces a more complex problem of scaling with a variable window where the data becomes positionally uncorrelated, and cannot be addressed by present methods.
- AI-driven analysis by identifying risk factors of VL relapse in HIV co-infected patients2
The results indicate the promise of explainable artificial intelligence (XAI) for making clinical decisions and remedying recurrences in VL and highlight the need to address the lack of transparency in Machine Learning (ML) models.
- packetLSTM: Dynamic LSTM Framework for Streaming Data with Varying Feature Space2
A dynamic LSTM-based novel method, called packetLSTM, to model the dimension-varying streams of streaming data and achieves state-of-the-art results on five datasets, and its underlying principle is extended to other RNN types, like GRU and vanilla RNN.
- GUNet++: guided-U-Net-based compact image representation with an improved reconstruction mechanism2
A deep-learning-based CIR technique that selects key pixels using the guided U-Net (GU-Net) architecture, and then near-original images are constructed using a conditional generative adversarial network (GAN)-based architecture is designed.
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- Explainable deep learning with Shapley additive explanations for multivariate wind power forecasting1
- 1
- 1
- MiShape: 3D Shape Modelling of Mitochondria in Microscopy1
MiShape is a generative model learned using implicit representations of mitochondrial shapes that provides a shape distribution that can be used to generate infinite realistic mitochondrial shapes and its utility for 3D shape reconstruction given a single 2D fluorescence image or a small 3D stack of 2D slices is demonstrated.
- 1
- 1
- On Designing Light-Weight Object Trackers Through Network Pruning: Use CNNS or Transformers?1
This work demonstrates how highly compressed light-weight object trackers can be designed using neural architectural pruning of large CNN and Transformer based trackers and provides deeper insights into designing highly efficient trackers from existing SOTA methods.
- Mabnet: Master Assistant Buddy Network With Hybrid Learning for Image Retrieval1
A novel Master Assistant Buddy Network (MAB-Net) for image retrieval which incorporates both the learning mechanisms and consists of master and assistant block, both learning independently through supervision and collectively via self-supervision.
- 1
- 1
- TMAV: Temporal Motionless Analysis of Video using CNN in MPSoC1
This work proposes a novel bio-inspired methodology, called IRON-MAN (Integrated Rational prediction and Motionless ANalysis), which utilizes Bayesian update on top of the individual image frame analysis in the videos and has resulted in highly accurate prediction of Temporal Motionless Analysis of the Videos (TMAV).
- 1
- Robust Ellipse Detection via Duality Principle with a False Determination Control1
A novel ellipse detection approach that eliminates false detection-based parameter space decomposition, principal of symmetric tangents, and a novel geometric constraint utilizing properties of tangents of ellipses is proposed.
- 1
- –
- A Modular Deep Learning Pipeline for Automated Sperm Motility Analysis–
A fully automated, open-source pipeline is presented that transforms raw phase-contrast microscopy video into WHO 2021 motility percentages and natural-language clinical reports with no manual intervention and confirms that standard WHO 2021 thresholds are empirical optimal for this imaging setup.
- –
- Live high-content imaging with automated analysis reveals mitochondrial changes during vascular calcification–
A High-throughput Mitochondrial Imaging Platform (H-MIP) combined with a deep learning segmentation model is presented, enabling rapid imaging and automated characterization of mitochondrial morphology in millions of mitochondria from tens of thousands of cells.
- –
- –
- –
- –
- Haphazard Inputs as Images in Online Learning–
This paper proposes to transform the varying feature space in an online learning setting to a fixed-dimension image representation on the fly, allowing any vision-based models to be applicable for haphazard inputs, as demonstrated using ResNet and ViT.
- –
- Graph-based hostile content detection in Hindi language–
This study proposes a scalable AI-based framework for detecting hostile posts in Hindi, the most widely spoken language in the Indian subcontinent and the third most spoken globally.
- –
- –
- Improving visual question answering for visually impaired people leveraging pre-trained models–
A pipeline that combines pre-processing and various information extraction techniques using pre-trained models, leading to an approximate 17% improvement in VQA accuracy compared to the best-performing single pre-trained model is proposed.
- –
- –
- –
- –
- –
- –
- –
- –
- Partial Binarization of Neural Networks for Budget-Aware Efficient Learning–
This paper proposes a controlled approach to partial binarization, creating a budgeted binary neural network (B2NN) with the MixBin strategy, which optimizes the mixing of binary and full-precision components, allowing for explicit selection of the fraction of the network to remain binary.
- –
- –
- –
- –
- –
- seMLP: Self-evolving Multi-layer Perceptron in Stock Trading Decision Making–
This paper proposes a novel self-evolving neural network system called seMLP which can abstract the data and produce an optimum neural network architecture without expert tuning and can automatically choose best performing models.
- –
- –
- –
- –
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-12. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.