A Survey on Human Detection using Reinforcement Learning

Human Detection - technology related to computer vision and image processing work by finding people in digital photos and videos and surveillance videos that are part of the observation. Single Shot Detector (SSD) is a deep learning method and is one of the fastest algorithms that use a single convolutional neural network to detect objects involving humans, cats, dogs, etc., and extract feature maps to classify the candidate object in the respective images. The advantage that SSD has is that it is quick to detect and has high accuracy in a given situation compared to regional suggested networks with smaller resolution images and smaller objects. However, it is still somewhat lagging in detecting large objects in larger images as compared to other algorithms that have been used to achieve better accuracy. It is a simple, end-to-end solution for a single network, and detection and extraction are done with one step forward single pass. The proposed system is to use the Optimized-SSD algorithm to detect human accuracy in the proposed database with good accuracy which will be the task of learning to increase SSD capacity as a detection system.

An Improved Approach for Object Proposals Generation

The objectness measure is a significant and effective method used for generic object detection. However, several object detection methods can achieve accurate results by using more than 1000 candidate object proposals. In addition, the weight of each proposal is weak and also cannot distinguish object proposals. These weak proposals have brought difficulties to the subsequent analysis. To significantly reduce the number of proposals, this paper presents an improved generic object detection approach, which predicts candidate object proposals from more than 10,000 proposals. All candidate proposals can be divided, rather than preclassified, into three categories: entire object, partial object, and nonobject. These partial object proposals also display fragmentary information of the objectness feature, which can be used to reconstruct the object boundary. By using partial objectness to enhance the weight of the entire object proposals, we removed a huge number of useless proposals and reduced the space occupied by the true positive object proposals. We designed a neural network with lightweight computation to cluster the most possible object proposals with rerank and box regression. Through joint training, the lightweight network can share the features with other subsequent tasks. The proposed method was validated using experiments with the PASCAL VOC2007 dataset. The results showed that the proposed approach was significantly improved compared with the existing methods and can accurately detect 92.3% of the objects by using less than 200 proposals.

Soft gamma-ray selected giant radio galaxies: an update

ABSTRACT We present an update on the sample of soft gamma-ray selected giant radio galaxies (GRGs) extracted from INTEGRAL/IBIS and Swift/BAT surveys; it includes eight new sources and one candidate object. In the new sample, all but one source display FR II radio morphologies; the only exception is B21144+35B, which is an FR I. The objects belong to both type 1 and type 2 active galactic nucleus (AGN) optical classes and have redshifts in the range 0.06–0.35, while the radio sizes span from 0.7 to 1 Mpc. In this study, we present for the first time two objects that were never discussed as GRGs before and propose a new candidate GRG. We confirm the correlation between the X-ray luminosity and the radio core luminosity found for other soft gamma-ray selected GRGs and expected for AGNs powered by efficient accretion. We also corroborate previous results that indicate that the luminosity of the radio lobes is relatively low compared with the nuclear X-ray emission. This supports the idea that the nucleus of these GRGs is now more powerful than in the past, consistent with a restarting activity scenario.

A hand-drawn map-based navigation method for mobile robots using objectness measure

Correct cognition of the environment is the premise of mobile robots to realize autonomous navigation control tasks. The inconsistency caused by time-varying environmental information is a bottleneck for the development and application of cognitive environment technologies. In this article, we propose an environmental cognition method that uses a hand-drawn map. Firstly, we use the single skeleton refinement and fuzzy c-means algorithms to segment the image. Then, we select candidate regions combining the saliency map. At the same time, we use the superpixels straddling method to filter the windows. The final candidate object regions are obtained based on a fusion of saliency segmentation and superpixels clustering. Based on the above objectness estimation results, we use a human–computer interaction method to construct an inaccurate hand-drawn environment map for navigation. The experimental results from PASCAL VOC2007 validate the efficacy of the proposed objectness measure method, where our result of 41.2% on mean average precision is the best of the tested methods. Furthermore, the experimental results of robot navigation in the actual scene also verified the effectiveness of the proposed approach.

Robust Scale Adaptive Visual Tracking with Correlation Filters

Visual tracking is a challenging task in computer vision due to various appearance changes of the target object. In recent years, correlation filter plays an important role in visual tracking and many state-of-the-art correlation filter based trackers are proposed in the literature. However, these trackers still have certain limitations. Most of existing trackers cannot well deal with scale variation, and they may easily drift to the background in the case of occlusion. To overcome the above problems, we propose a Correlation Filters based Scale Adaptive (CFSA) visual tracker. In the tracker, a modified EdgeBoxes generator, is proposed to generate high-quality candidate object proposals for tracking. The pool of generated candidate object proposals is adopted to estimate the position of the target object using a kernelized correlation filter based tracker with HOG and color naming features. In order to deal with changes in target scale, a scale estimation method is proposed by combining the water flow driven MBD (minimum barrier distance) algorithm with the estimated position. Furthermore, an online updating schema is adopted to reduce the interference of the surrounding background. Experimental results on two large benchmark datasets demonstrate that the CFSA tracker achieves favorable performance compared with the state-of-the-art trackers.

Low-level aspects of segmentation and recognition

This paper discusses two problems related to three-dimensional object recognition. The first is segmentation and the selection of a candidate object in the image, the second is the recognition of a three-dimensional object from different viewing positions. Regarding segmentation, it is shown how globally salient structures can be extracted from a contour image based on geometrical attributes, including smoothness and contour length. This computation is performed by a parallel network of locally connected neuron-like elements. With respect to the effect of viewing, it is shown how the problem can be overcome by using the linear combinations of a small number of two-dimensional object views. In both problems the emphasis is on methods that are relatively low level in nature. Segmentation is performed using a bottom -up process, driven by the geometry of image contours. Recognition is performed without using explicit three-dimensional models, but by the direct manipulation of two-dimensional images.