A5022393071- 3D Surveying and Cultural Heritage
- 3D Shape Modeling and Analysis
- Advanced Image Processing Techniques
- Image Enhancement Techniques
- Advanced Image and Video Retrieval Techniques
- Advanced Image Fusion Techniques
- Advanced Vision and Imaging
- Domain Adaptation and Few-Shot Learning
- Computer Graphics and Visualization Techniques
- Image Processing and 3D Reconstruction
- Remote Sensing and LiDAR Applications
- Image and Signal Denoising Methods
- Human Pose and Action Recognition
- Advanced Optical Sensing Technologies
- Anomaly Detection Techniques and Applications
- Image Processing Techniques and Applications
- Robotics and Sensor-Based Localization
- Advanced Neural Network Applications
- Infrared Target Detection Methodologies
- Remote-Sensing Image Classification
- Maritime Navigation and Safety
- Advanced Numerical Analysis Techniques
- Image Retrieval and Classification Techniques
- Multimodal Machine Learning Applications
- Video Surveillance and Tracking Methods
KLE Technological University
2017-2024
This paper reviews the NTIRE 2022 challenge on night photography rendering. The solicited solutions that processed RAW camera images captured in scenes to produce a photo-finished output image encoded standard RGB (sRGB) space. Given subjective nature of this task, proposed were evaluated based mean opinions viewers asked judge visual appearance results. Michael Freeman, world-renowned photographer, further ranked with highest opinion scores. A total 13 teams competed final phase challenge....
The 1 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">st</sup> Workshop on Maritime Computer Vision (MaCVi) 2023 focused maritime computer vision for Unmanned Aerial Vehicles (UAV) and Surface Vehicle (USV), organized several subchallenges in this domain: (i) UAV-based Object Detection, (ii) Mar-itime Tracking, (iii) USV-based Obstacle Segmentation (iv) Detection. were based the SeaDronesSee MODS benchmarks. This report summarizes main findings...
In this paper, we propose a generative model to restore degraded underwater images considering attenuation coefficients as clue and name it AquaGAN. Computing the given in revised image formation demands in-situ measurements. However, measurements scenario is infeasible. Towards this, estimate using learning based methods use these parameters for restoration of images. Restoration true colors challenging intensity light changes with distance. Preserving during by minimizing single objective...
In this paper, we propose a learning-based approach for automatic detection of hole boundary points in 3D point cloud. Point cloud is an important geometric data structure used modelling. Data obtained from acquisition techniques often result deficiencies such as holes the For successful hole-filling to achieve better surface reconstruction, accurate necessary. Most existing methods use threshold values different parameters which need be set manually after analyzing nature It becomes...
In this paper, we present a novel method for generating synthetic underwater images considering revised image formation model. We propose to use the generated train conditional generative adversarial network (CGAN) towards restoration of degraded images. Restoration using traditional dehazing models is challenging as they are insensitive wavelength, depth, water type and treat backscattering direct signal attenuation coefficients be equal. However, learning based perform well but sensitive...
In this paper, we propose VG-VAE: Venatus Geometric Variational Auto-Encoder for capturing unsupervised hierarchical local and global geometric signatures in pointcloud. Recent research emphasises the significance of underlying intrinsic geometry pointcloud processing. Our contribution is to extract analyse morphology using proposed Proximity Correlator (GPC) variational sampling latent. The extraction facilitated by GPC, whereas sampling. Furthermore, apply a naive mix vector algebra 3D...
Developing and integrating advanced image sensors with novel algorithms in camera systems are prevalent the increasing demand for computational photography imaging on mobile platforms. However, lack of high-quality data research rare opportunity in-depth exchange views from industry academia constrain development intelligent (MIPI). With success 1st MIPI Workshop@ECCV 2022, we introduce second challenge including four tracks focusing algorithms. In this paper, summarize review Nighttime...
In this paper, we propose DeFi: a novel perspective for hole detection and filling of given deteriorated 3D point cloud towards digital preservation cultural heritage sites. Preservation demands digitization as sites deteriorate due to natural calamities human activities. Digital promotes acquisition data using sensor or Multi-view reconstruction. Unfortunately, finds challenges the limitations in technology inappropriate capture conditions, leading formation missing regions holes acquired...
Underwater images suffer from haze, blue/green tint, and color distortion, caused by attenuation scattering of light. Existing methods for restoration depend on transmission maps to inverse the effects based underwater image formation model, enhancement often result in unnatural colors. In this paper, we propose a method adaptively estimate correction curve CIE L*a*b* space single enhancement. Haze blue-green tint are removed through process. We demonstrate improved canny edge detection...
In this paper, we perform restoration of underwater images by considering principles the image formation model in deep neural networks. Typically, suffer from blur, color loss and other degradations due to scattering absorption light water as a medium. Quality is sensitive depth increases with introduces considerable amount degradation. However, literature infer, recent frameworks do not consider influence on under-water images. Towards this, propose clue for relative distance objects scene....
In this paper, we propose a framework for enhancement of underwater images. Underwater images suffer from low-contrast, blur and non-uniform illumination resulting in poor quality Red color the atmospheric light is absorbed early due to its shorter wavelength, whereas colors like blue green penetrate deeper into water larger wavelength. As result appear bluish or greenish color. Towards this, using balance Laplacian Gaussian fusion pyramid. Here aim distribution image LAB space, remove...
In this paper, we propose a Deep Dense Network for Depth Completion Task (DeepDNet) towards generating dense depth map using sparse and captured view. Wide variety of scene understanding applications such as 3D reconstruction, mixed reality, robotics demand accurate maps. Existing sensors capture reliable find challenges in acquiring Towards plan to utilise the input with RGB image generate depth. We model transformation random grid-based Quad-tree decomposition. Dense-Residual-Skip (DRS)...
In this paper, we address the problem of decision fusion for robust horizon estimation using Dempster Shafer Combination Rule (DSCR). We provide a framework to select estimate out 'n' estimates, based on confidence factor. Vision-based attitude depends and no single algorithm gives accurate results different kind scenarios. propose combine evidence parameters generate factor DSCR justify correctness estimated horizon. compute Confidence Interval (CI) Gaussian Mixture Model (GMM). also two...
In this work, we propose a generative model for enhancement of images captured in low-light conditions. Sensor constraints and inappropriate lighting conditions are accountable degradations introduced the image. The limit visibility scene impedes vision applications like detection, tracking surveillance. Recently, deep learning algorithms have taken leap However, these fail to capture information on fine grained local structures performance. Towards this, low-lit exploit both global...
In this paper, we propose multilevel framework for summarization of surveillance videos using motion entropy by maintaining chronology activities. We aim to reduce the size video give a meaningful summary retaining important activities without destroying temporal relationship Initially input is divided into blocks and then segments in with non-uniform number frames block at each level. present mechanism select salient which can contribute propagating information from frame level segment...
In this paper, we propose a framework to categorize heritage site images based on unsupervised clustering for 3D reconstruction. Modeling generalized classification model crowdsourced data is challenging as the invariant and high dimensional. Handcrafted features find challenges in differentiating such data. To address this, cluster using deep features. We Inception Variational Autoencoder(IVAE) extracting 8 from an manner. Towards refining new loss function, variant of Autoencoder (VAE)...
In this paper, we address the problem of 3D Point Cloud Upsampling, that is, given a set points, objective is to obtain denser point cloud representation. We achieve by proposing deep learning architecture along with consuming clouds directly, also accepts associated auxiliary information such as Normals and Colors consequently upsamples them. design novel feature loss function train model. demonstrate our work on ModelNet dataset show consistent improvements over existing methods.
In this paper, we propose IPD-Net: Invariant Primitive Decompositional Network, a SO(3) invariant framework for decomposition of point cloud. The human cognitive system is able to identify and interpret familiar objects regardless their orientation abstraction. Recent research aims bring capability machines understanding the 3D world. work, present inspired by cognition decompose clouds into four primitive shapes (plane, cylinder, cone, sphere) enable understand irrespective its...
In this paper, we propose TP-NoDe, a novel Topology-aware Progressive Noising and Denoising technique for 3D point cloud upsampling. TP-NoDe revisits the traditional method of upsampling by introducing perspective adding local topological noise incorporating algorithm Density-Aware k nearest neighbour (DA-kNN) followed denoising to map noisy perturbations topology cloud. Unlike previous methods, progressively upsample cloud, starting at 2 × ratio advancing desired ratio. generates...
The 1$^{\text{st}}$ Workshop on Maritime Computer Vision (MaCVi) 2023 focused maritime computer vision for Unmanned Aerial Vehicles (UAV) and Surface Vehicle (USV), organized several subchallenges in this domain: (i) UAV-based Object Detection, (ii) Tracking, (iii) USV-based Obstacle Segmentation (iv) Detection. were based the SeaDronesSee MODS benchmarks. This report summarizes main findings of individual introduces a new benchmark, called Detection v2, which extends previous benchmark by...