Search results for "Object Detection"
showing 10 items of 64 documents
Context-Aware Model Applied to Hog Descriptor for People Detection
2018
International audience; This work proposes and implements a method based on Context-Aware Visual Attention Model (CAVAM), but modifying the method in such way that the detection algorithm is replaced by Histograms of Oriented Gradients (HOG). After reviewing different algorithms for people detection, we select HOG method because it is a very well known algorithm, which is used as a reference in virtually all current research studies about automatic detection. In addition, it produces accurate results in significantly less time than many algorithms. In this way, we show that CAVAM model can be adapted to other methods for object detection besides Scale-Invariant Feature Transform (SIFT), as …
Estimating Tree Health Decline Caused by Ips typographus L. from UAS RGB Images Using a Deep One-Stage Object Detection Neural Network
2022
Various biotic and abiotic stresses are causing decline in forest health globally. Presently, one of the major biotic stress agents in Europe is the European spruce bark beetle (Ips typographus L.) which is increasingly causing widespread tree mortality in northern latitudes as a consequence of the warming climate. Remote sensing using unoccupied aerial systems (UAS) together with evolving machine learning techniques provide a powerful tool for fast-response monitoring of forest health. The aim of this study was to investigate the performance of a deep one-stage object detection neural network in the detection of damage by I. typographus in Norway spruce trees using UAS RGB images. A Scaled…
Towards General Purpose Object Detection: Deep Dense Grid Based Object Detection
2020
Object detection is one of the most challenging and very important branch of computer vision. Some of the challenging aspect of a detection network is the fact that an object can appear anywhere in the image, be partially occluded by another object, might appear in crowd or have greatly varying scales. Consequently, we propose a fine grained and equally spaced dense grid cells throughout an input image be responsible of detecting an object. We re-purpose an already existing deep state-of-the-art detector or classifier into deep and dense detector. Our dense object detector uses binary class encoding and hence suitable for very large multi-class object detector. We also propose a more flexib…
Three-dimensional object detection under arbitrary lighting conditions
2006
A novel method of 3D object recognition independent of lighting conditions is presented. The recognition model is based on a vector space representation using an orthonormal basis generated by the Lambertian reflectance functions obtained with distant light sources. Changing the lighting conditions corresponds to multiplying the elementary images by a constant factor and because of that, all possible lighting views will be elements that belong to that vector space. The recognition method proposed is based on the calculation of the angle between the vector associated with a certain illuminated 3D object and that subspace. We define the angle in terms of linear correlations to get shift and i…
Automatic object detection in point clouds based on knowledge guided algorithms
2013
The modeling of real-world scenarios through capturing 3D digital data has been proven applicable in a variety of industrial applications, ranging from security, to robotics and to fields in the medical sciences. These different scenarios, along with variable conditions, present a challenge in discovering flexible appropriate solutions. In this paper, we present a novel approach based on a human cognition model to guide processing. Our method turns traditional data-driven processing into a new strategy based on a semantic knowledge system. Robust and adaptive methods for object extraction and identification are modeled in a knowledge domain, which has been created by purely numerical strate…
The iterative object symmetry transform
2005
This paper introduces a new operator named the Iterated Object Transform that is computed by combining the Object Symmetry Transform with the morphological operator erosion. This new operator has been applied on both binary and gray levels images showing the ability to grasp the internal structure of a digital object. We present some experiments on real images in face analysis.
A hardware skin-segmentation IP for vision based smart ADAS through an FPGA prototyping
2017
International audience; In this paper we presents a platform based design approach for fast HW/SW embedded smart Advanced Driver Assistant System (ADAS) design and prototyping. Then, we share our experience in designing and prototyping a HW/SW vision based smart embedded system as an ADAS that helps to increase the safety of car's drivers. We present a physical prototype of the vision ADAS based on a Zynq FPGA. The system detects the fatigue state of the driver by monitoring the eyes closure and generates a real-time alert. A new HW/SW codesign skin segmentation step to locate the eyes/face is proposed. Our presented new approach migrates the skin segmentation step from processing system (S…
Scratch detection and removal from static images using simple statistics and genetic algorithms
2002
This paper investigates the removal of line scratches from old movies and gives a twofold contribution. First, it presents simple technique for detecting the scratches, based on an analysis of the statistics of the grey levels. Second, the scratch removal is approached as an optimisation problem, and it is solved by using a genetic algorithm. The method can be classified as a static approach, as it works independently on each single frame of the sequence. It does not require any a-priori knowledge of the absolute position of the scratch, nor an external starting population of chromosomes for the genetic algorithm. The central column of the line scratch once detected is changed with a conven…
Zero-shot Semantic Segmentation using Relation Network
2021
Zero-shot learning (ZSL) is widely studied in recent years to solve the problem of lacking annotations. Currently, most studies on ZSL are for image classification and object detection. But, zero-shot semantic segmentation, pixel level classification, is still at its early stage. Therefore, this work proposes to extend a zero-shot image classification model, Relation Network (RN), to semantic segmentation tasks. We modified the structure of RN based on other state-of-the-arts semantic segmentation models (i.e. U-Net and DeepLab) and utilizes word embeddings from Caltech-UCSD Birds 200-2011 attributes and natural language processing models (i.e. word2vec and fastText). Because meta-learning …
Editorial for the Special Issue “Frontiers in Spectral Imaging and 3D Technologies for Geospatial Solutions”
2019
This Special Issue hosts papers on the integrated use of spectral imaging and 3D technologies in remote sensing, including novel sensors, evolving machine learning technologies for data analysis, and the utilization of these technologies in a variety of geospatial applications. The presented results showed improved results when multimodal data was used in object analysis.