A5102782582- Remote Sensing in Agriculture
- Remote-Sensing Image Classification
- Advanced Neural Network Applications
- Emotion and Mood Recognition
- Remote Sensing and LiDAR Applications
- Video Surveillance and Tracking Methods
- Human Pose and Action Recognition
- Land Use and Ecosystem Services
- EEG and Brain-Computer Interfaces
- Domain Adaptation and Few-Shot Learning
- Advanced Image and Video Retrieval Techniques
- Automated Road and Building Extraction
- Video Analysis and Summarization
- Color perception and design
- Anomaly Detection Techniques and Applications
- Quantum Computing Algorithms and Architecture
- Generative Adversarial Networks and Image Synthesis
- Geochemistry and Geologic Mapping
- Multimodal Machine Learning Applications
- Cryptography and Data Security
- Handwritten Text Recognition Techniques
- Medical Image Segmentation Techniques
- Quantum Information and Cryptography
- Face and Expression Recognition
- Context-Aware Activity Recognition Systems
Maulana Abul Kalam Azad University of Technology, West Bengal
2015-2025
Variable Energy Cyclotron Centre
2024
Scuola Superiore Sant'Anna
2024
Homi Bhabha National Institute
2024
Maulana Abul Kalam Azad Institute of Asian Studies
2021-2022
Indian Institute of Technology Kharagpur
2017-2022
Meta (Israel)
2018-2021
Louisiana State University
2012-2019
Louisiana State University Agricultural Center
2016
National Institute of Technology Durgapur
2009-2011
We present the DeepGlobe 2018 Satellite Image Understanding Challenge, which includes three public competitions for segmentation, detection, and classification tasks on satellite images (Figure 1). Similar to other challenges in computer vision domain such as DAVIS[21] COCO[33], proposes datasets corresponding evaluation methodologies, coherently bundled with a dedicated workshop co-located CVPR 2018. observed that imagery is rich structured source of information, yet it less investigated...
Satellite image classification is a challenging problem that lies at the crossroads of remote sensing, computer vision, and machine learning. Due to high variability inherent in satellite data, most current object approaches are not suitable for handling datasets. The progress analytics has also been inhibited by lack single labeled high-resolution dataset with multiple class labels. contributions this paper twofold -- (1) first, we present two new datasets called SAT-4 SAT-6, (2) then,...
Road network extraction from satellite images often produce fragmented road segments leading to maps unfit for real applications. Pixel-wise classification fails predict topologically correct and connected masks due the absence of connectivity supervision difficulty in enforcing topological constraints. In this paper, we propose a task called Orientation Learning, motivated by human behavior annotating roads tracing it at specific orientation. We also develop stacked multi-branch...
Recognition of emotion is always a difficult problem, particularly if the recognition done by using speech signal. Many significant research works have been on The primary challenges are choosing corpora (speech database), identification different features related to and an appropriate choice classification model. In this article we use 13 MFCC (Mel Frequency Cepstral Coefficient) with velocity acceleration component as CNN (Convolution Neural Network) LSTM (Long Short Term Memory) based...
Quantum annealing is an experimental and potentially breakthrough computational technology for handling hard optimization problems, including problems of computer vision. We present a case study in training production-scale classifier tree cover remote sensing imagery, using early-generation quantum hardware built by D-wave Systems, Inc. Beginning within known boosting framework, we train decision stumps on texture features vegetation indices extracted from four-band, one-meter-resolution...
Satellite image classification is a challenging problem that lies at the crossroads of remote sensing, computer vision, and machine learning. Due to high variability inherent in satellite data, most current object approaches are not suitable for handling datasets. The progress analytics has also been inhibited by lack single labeled high-resolution dataset with multiple class labels. contributions this paper twofold - (1) first, we present two new datasets called SAT-4 SAT-6, (2) then,...
The use of satellite imagery has become increasingly popular for disaster monitoring and response. After a disaster, it is important to prioritize rescue operations, response coordinate relief efforts. These have be carried out in fast efficient manner since resources are often limited disaster-affected areas it's extremely identify the maximum damage. However, most existing mapping efforts manual which time-consuming leads erroneous results. In order address these issues, we propose...
Accurate tree-cover estimates are useful in deriving above-ground biomass density from very high resolution (VHR) satellite imagery data. Numerous algorithms have been designed to perform delineation high-to-coarse-resolution imagery, but most of them do not scale terabytes data, typical these VHR data sets. In this paper, we present an automated probabilistic framework for the segmentation and classification 1-m as obtained National Agriculture Imagery Program (NAIP) whole Continental...
This paper considers two dimensional valence-arousal model. Pictorial stimuli of International Affective Picture Systems were chosen for emotion elicitation. Physiological signals like, Galvanic Skin Response, Heart Rate, Respiration Rate and Temperature measured accessing emotional responses. The experimental procedure uses non-invasive sensors signal collection. A group healthy volunteers was shown four types categorized as High Valence Arousal, Low Arousal around thirty minutes Linear...
We investigate the use of Deep Neural Networks for classification image datasets where texture features are important generating class-conditional discriminative representations. To this end, we first derive size feature space some standard textural extracted from input dataset and then theory Vapnik-Chervonenkis dimension to show that hand-crafted extraction creates low-dimensional representations which help in reducing overall excess error rate. As a corollary analysis, time upper bounds...
Satellite image classification is a challenging problem that lies at the crossroads of remote sensing, computer vision, and machine learning. Due to high variability inherent in satellite data, most current object approaches are not suitable for handling datasets. The progress analytics has also been inhibited by lack single labeled high-resolution dataset with multiple class labels. In preliminary version this work, we introduced two new resolution imagery datasets (SAT-4 SAT-6) proposed...
Emotions and affect are universal means of expressing the physiological state an individual. Most our daily interactions with other individuals involve emotions as integral part. It has become highly prominent in technological research, new technologies related to human-machine interaction or medical applications developed. Detecting analyzing have quite important area research. In this paper, we took effort find effect different on signals. We used International Affective Picture System...
Abstract. In this paper, we perform multi-sensor multi-resolution data fusion of Landsat-5 TM bands (at 30 m spatial resolution) and multispectral World View-2 (WV-2 at 2 through linear spectral unmixing model. The advantages fusing Landsat WV-2 are two fold: first, resolution the increases to resolution. Second, integration from sensors allows additional SWIR fused product which have such as improved atmospheric transparency material identification, for example, urban features, construction...
Consider a situation in which the transmission of encrypted message is intercepted by an adversary who can later ask sender to reveal random choices (and also secret key, if one exists) used generating cipher text, thereby exposing plaintext. An encryption scheme deniable generate `fake choice' that will make text `look like' different plaintext, thus keeping real plaintext private. Analogous requirements be formulated with respect attacking receiver and both parties. In this paper we...
Classification techniques for images of handwritten characters are susceptible to noise. Quadtrees can be an efficient representation learning from sparse features. In this paper, we improve the effectiveness probabilistic quadtrees by using a pixel level classifier extract character pixels and remove noise images. The denoiser (a deep belief network) uses map responses obtained pretrained CNN as features reconstructing eliminating We experimentally demonstrate our approach classifying noisy...