Identification Of Target Objects Research Articles

Win-Win Cooperation: Semantic Encoding Learning and Saliency Selection for Weakly supervised Semantic Segmentation

Image-level weakly supervised semantic segmentation methods have attracted increasing attention due to data labeling efficiency, but these methods mostly focus on utilizing the localization maps generated by the classification network to produce pseudo labels, leading to sparse object regions, object boundary mismatch and co-occurring pixels existing in the target objects. To address these issues, we propose a novel image-level weakly supervised semantic segmentation algorithm, namely Semantic Encoding Learning and Saliency Selection (SELSS), which mainly focuses on the improvement for target object identification and boundary quality. Specifically, we design a semantic encoding learning module to help the localization map from the classification network cover more semantic regions, which measures the euclidean distance between semantic words and localization maps to obtain the object coverage identification. In order to obtain accurate object boundaries and discard co-occurring pixels, we utilize the encoded localization maps for the foreground and the background to perform the saliency selection under the pseudo-pixel feedback. Under the cooperation between the semantic encoding learning and the saliency selection, our SELSS can better tackle the key challenges existing in weakly supervised semantic segmentation, significantly improving the quality of the generated pseudo labels. Extensive experiments demonstrate that our SELSS method achieves the state-of-the-art performance on the PASCAL VOC 2012 and MS COCO 2014 segmentation benchmarks.

Simple Non‐Destructive and 3D Multi‐Layer Visual Hull Reconstruction with an Ultrabroadband Carbon Nanotubes Photo‐Imager

AbstractWhile millimeter‐wave (MMW)–infrared (IR) broad photo‐imaging exhibits significant potential as an inspection technique, photo‐thermoelectric (PTE) detector operations employing carbon nanotubes (CNTs) are highly suitable because of their efficient absorption characteristics in the above bands. Such an imaging device design and the associated material coordination allow the aggregation of multiple wavelength‐specific optical information for the constituent identification of target objects in a non‐destructive manner with high sensitivity. On the other side, computer vision techniques facilitate 3D structure reconstruction as approaches from inspection methodologies, in addition to the aforementioned imaging device design. Although non‐destructive inspections require collectively satisfying both constituent identification and structural reconstruction, investigations combining MMW–IR broad photo imaging and computer vision monitoring are still insufficient. This work demonstrates computer vision‐driven simple 3D composite multi‐layer reconstructions via non‐destructive broadband multiple‐wavelength monitoring with the CNT film PTE imager. Visual hull evaluation, which is one of the representative computer vision techniques, converts 2D transmissive PTE images with different viewpoints into 3D reconstruction models. The subsequential graphical superposition of these wavelength‐specific models results in non‐destructive entire reconstructions of 3D composite multi‐layer target objects. Therefore, the present computer vision‐driven PTE broadband 3D reconstruction bridges the gap between methodologies and device design strategies toward non‐destructive inspection applications.

Виртуальные среды для моделирования взаимодействия операторов с БПЛА в закрытых пространствах в потенциально опасных ситуациях

The effective interaction between humans and robotic devices is designed to provide a timely response to potentially dangerous situations arising in complex environments. One example is the use of Unmanned Aerial Vehicles (UAVs) by human operators to inspect enclosed spaces. These units allow the search and identification of target objects, helping a human operator identify signs of Potentially Dangerous Situations (PDSs). At the same time in a closed (and often cluttered) space to control the emerging situations it is necessary to involve developed means of conducting spatial orientation based on Artificial Intelligence Technology, Computer Vision, 3D-visualization, Local Positioning And Navigation, etc. Despite the progress made in autonomous UAVs control, active human participation in detecting signs of PDSs is an important condition for maintaining the security of the internal environment of controlled premises. Human decision-making in situations of high uncertainty and incomplete data largely depends on means to improve his Situational Awareness (external conditions of activity) and Professional Competencies (internal conditions of activity). To form them, it is necessary to develop methodology and tools for building UAVs application scenarios to give operators the necessary user experience on simulation models in the Virtual Environment of Activity. The paper investigated the development of the Virtual Environment Systems (VES) for the implementation of UAVs application scenarios in enclosed spaces in relation to multicopters that allow both autonomous and manual control. As a result of this research, prototypes of UAVs for inspection of enclosed spaces have been identified; the prospects of quadcopters use under various PDSs scenarios have been shown; the composition of promising technologies ensuring UAVs use in enclosed environments has been described; the method of design and 3D visualization of UAVs application scenarios using the original VES (named VirSim) has been proposed. The advantages of the proposed approach to modeling the use of UAVs in PDSs are the multivariate scenarios of operators' activities, which contributes to the accumulation of user experience of human interaction with flying robots.

Pipe pose estimation based on machine vision

To realize the automatic assembly and connection of pipelines, one of the core tasks is target object identification and pose estimation. The problem is challenging due to low precision and efficiency caused by the pipes being textureless and self-occluding. In this work, we introduce a machine vision-based method for 6D pipe pose estimation. First, the pipe’s initial pose is estimated by a template matching algorithm. Second, a 3D-2D projection mapping relationship is established, and the distance between the edge pixel points and the edge of the projection model is optimized using the least squares method to obtain a more accurate pipe pose. Experiments demonstrate that the proposed method is able to robustly estimate pose in real environments, while achieving position and pitch accuracies of 0.0732 mm and 0.5°, respectively. Furthermore, the whole pose estimation process lasted 2–3 s, which meets the requirements of industrial applications.

Identifying the vegetation type in Google Earth images using a convolutional neural network: a case study for Japanese bamboo forests

BackgroundClassifying and mapping vegetation are crucial tasks in environmental science and natural resource management. However, these tasks are difficult because conventional methods such as field surveys are highly labor-intensive. Identification of target objects from visual data using computer techniques is one of the most promising techniques to reduce the costs and labor for vegetation mapping. Although deep learning and convolutional neural networks (CNNs) have become a new solution for image recognition and classification recently, in general, detection of ambiguous objects such as vegetation is still difficult. In this study, we investigated the effectiveness of adopting the chopped picture method, a recently described protocol for CNNs, and evaluated the efficiency of CNN for plant community detection from Google Earth images.ResultsWe selected bamboo forests as the target and obtained Google Earth images from three regions in Japan. By applying CNN, the best trained model correctly detected over 90% of the targets. Our results showed that the identification accuracy of CNN is higher than that of conventional machine learning methods.ConclusionsOur results demonstrated that CNN and the chopped picture method are potentially powerful tools for high-accuracy automated detection and mapping of vegetation.

Material Identification with Cosmic Ray Muons using RPCs

We plan to build an imaging setup for material identification utilizing the Coulomb scattering of cosmic ray muons due to their interaction with the materials and tracking their trajectories with Resistive Plate Chambers (RPCs). To begin with, we consider a setup of six RPCs stacked in a parallel manner to read the position and timing information of the muons before and after their interaction with a phantom of a given material using a set of three RPCs for each phase. Here we present a simulation work carried out to study the image formation of phantoms of several materials. A detailed modeling of the imaging system consisting of six RPCs was done using GEANT4. Cosmic Ray Library (CRY) was used for generation of particles with the appropriate energy and zenith angle distribution. Three reconstruction algorithms were followed for material identification and image reconstruction, viz. Point of Closest Approach (POCA), Iterative POCA and the Binned Cluster Algorithm. A weighted metric discriminator was calculated for target object identification. Using the algorithms, the imaging of the Region of Interest (ROI) lying between the two layers of RPCs was done. The time required to discriminate target objects and do the image reconstruction has been studied.

A fast video haze removal algorithm via mixed transmissivity optimisation

Image restoration is an important approach to image and video defogging. One of the most popular algorithms for image restoration is dark channel prior. However, when the algorithm is applied to outdoor digital webcams with limited computing resources, its real-time performance probably cannot be guaranteed. To address the above issue, this paper presents a fast video haze removal algorithm based on mixed optimised transmissivity to improve the time performance of the dark channel prior algorithm. The proposed algorithm combines guided filter and median filter and replaces the soft matting procedure in the classical dark channel prior algorithm. A set of experiments are performed to evaluate the real-time performance and effectiveness of the algorithm. The results show that our proposed improved algorithm can significantly improve the speed of video defogging, without sacrificing much effectiveness in identification of target objects.

A fast video haze removal algorithm via mixed transmissivity optimisation

Image restoration is an important approach to image and video defogging. One of the most popular algorithms for image restoration is dark channel prior. However, when the algorithm is applied to outdoor digital webcams with limited computing resources, its real-time performance probably cannot be guaranteed. To address the above issue, this paper presents a fast video haze removal algorithm based on mixed optimised transmissivity to improve the time performance of the dark channel prior algorithm. The proposed algorithm combines guided filter and median filter and replaces the soft matting procedure in the classical dark channel prior algorithm. A set of experiments are performed to evaluate the real-time performance and effectiveness of the algorithm. The results show that our proposed improved algorithm can significantly improve the speed of video defogging, without sacrificing much effectiveness in identification of target objects.

Not Only Whites: Racial Priming Effect for Black Faces in Black People

Black people are widely negatively stereotyped. The presence of unconscious stereotypes can be effectively assessed with the administration of “racial priming tasks.” An ethnically diverse group was subjected to a priming paradigm to test whether racial cues could bias the identification of target objects. Participants were asked to categorize objects (either as dangerous or nondangerous) after the presentation of Black/White faces as primes. Results show that both Black and White participants were faster in categorizing dangerous objects when primed with Black faces compared to the control condition (i.e., scrambled faces). One possible explanation for this effect is that Black faces are generally associated with a feeling of danger, which ultimately leads to faster responses.

Are there benefits of Visual Crowding?

Objects in peripheral vision are less identifiable when flanked by nearby distractors. This loss in performance, known as visual crowding, has traditionally been viewed to result from a bottleneck in visual processing (e.g. Levi, Vision Research, 2008). Here we will present the alternate view that crowding could serve to suppress the representation of repetitive items in peripheral vision so that attentional resources can be directed to non-repetitive, interesting and novel items, potentially as the targets of upcoming saccades. Traditionally, crowding experiments have maximized the degradation of target-object identification by presenting targets and distractors with similar characteristics (Kooi et al., Spatial Vision, 1994) or that group together strongly (Manassi et al., Journal of Vision, 2012). These conditions yield robust target-flanker interactions extending over long distances (Bouma, Nature 1970; Tripathy et al., Journal of Vision, 2014). However, rather than representing a limitation of peripheral visual processing, the strong suppression of repetitive or similar items might be desirable, serving to enhance the individuation and processing of eccentrically viewed novel objects. In agreement with this interpretation, crowding is known to be minimal in foveal vision, is strongest between repetitive or similar objects in peripheral vision and between items that readily group together, and is reduced by target-flanker differences in contrast polarity, color, depth, shape, or complexity. Crowding also is weakened when transients increase the conspicuity of the target relative to the flankers and when the target represents the end-point of a planned saccade. This presentation will evaluate previous studies of crowding to ascertain whether the results can be understood better in terms of an implicit deficit in the processing of nearby visual targets or in terms of a beneficial suppression of repetitive peripheral stimuli so that processing resources can be devoted more fully to unfixated objects that are novel and potentially interesting. Meeting abstract presented at VSS 2017

An Investigation on Effects of Harmattan Dust (aerosols) on Horizontal Visibility Deterioration over Bauchi, North- eastern Nigeria

A B S T R A C T In this paper, the effects of harmattan dust (aerosols) on visibility deterioration over Bauchi, a town in North-eastern Nigeria was investigated. The data of visibility readings for a period of ten (10) years were obtained by human identification of target objects and landmarks, most especially hills and buildings at known distances in a full 360 degrees circle around the observation point. The annual cycles of visibility in Bauchi for daily and monthly averages for ten years (1998-2007) were obtained. This enabled us to determine atmospheric extinction coefficient (βext.), which results in the scattering and absorption due to dust particles by using the Koschmieder relationship for the study area. Bauchi recorded a maximum visibility value of about 35000.0m in April/May 2005. The minimum visibility value of nearly zero was recorded on November/December 2000. However, during the months of June-September, the maximum atmospheric extinction coefficient value of 0.230 was recorded in 2007 and the minimum atmospheric extinction coefficient value of 0.164 was recorded in 1999. For the months of harmattan season (November to February), the maximum atmospheric extinction coefficient value of 0.689 was recorded in 2000/2001 and the minimum atmospheric extinction coefficient value of 0.270 was recorded in 2005/2006.

Target Object Identification and Location Based on Multi-sensor Fusion

For an unknown environment, how to make a mobile robot identify a target object and locate it autonomously, this is a very challenging question. In this paper, a novel multi-sensor fusion method based on a camera and a laser range finder (LRF) for mobile manipulations is proposed. Although a camera can acquire large quantities of information, it does not directly get the 3D data of the environment. Moreover, the camera image processing is complex and easily influenced from the change in ambient light. In view of the ability of the LRF to directly get the 3D coordinates of the environment and its stability against outside influence, and the superiority of the camera to acquire rich color information, the combination of the two sensors by making use of their advantages is employed to obtain more accurate measurement as well as to simplify information processing. To overlay the camera image with the measurement point cloud of the pitching LRF and to reconstruct the 3D image which includes pixel depth information, the homogeneous transformation model of the system is built. Then, based on the combination of the color features from the camera image and the shape features from the LRF measurement data, the autonomous identification and location of target object are achieved. In order to extract the shape features of the object, a two-step method is introduced, and a sliced point cloud algorithm is proposed for the preliminary classification of the measurement data of the LRF . The effectiveness of the proposed method is validated by the experimental testing and analysis carried out on the mobile manipulator platform. The experimental results show that by this method, the robot can not only identify target object autonomously, but also determine whether it can be operated, and acquire a proper grasping location.

On the relations between action planning, object identification, and motor representations of observed actions and objects

Recent evidence suggests that viewing a static prime object (a hand grasp), can activate action representations that affect the subsequent identification of graspable target objects. The present study explored whether stronger effects on target object identification would occur when the prime object (a hand grasp) was made more action-rich and dynamic. Of additional interest was whether this type of action prime would affect the generation of motor activity normally elicited by the target object. Three experiments demonstrated that grasp observation improved the identification of grasp-congruent target objects relative to grasp-incongruent target objects. We argue from this data that identifying a graspable object includes the processing of its action-related attributes. In addition, grasp observation was shown to influence the motor activity elicited by the target object, demonstrating interplay between action-based and object-based motor coding.

The applicability of national forest inventories for estimating forest damage outbreaks – Experiences from a Gremmeniella outbreak in Sweden

The design of the Swedish National Forest Inventory as well as the National Forest Damage Inventory is a sparse sample of systematically allocated plots. In this study data were combined from these two independent inventories to estimate geographical distribution, area extension, and disease development of a Gremmeniella abietina (Lagerb.) Morelet epidemic. For the combined data the standard error for estimated total area of affected Scots pine (Pinus sylvestris L.) and lodgepole pine (Pinus contorta Dougl. ex Loud. var.latifolia Engelm.) forests was about 11%. Assessments of the proportion of pine trees with fresh shoot blight infection shows that changes larger than 1% are significantly (p < 0.05) estimated. By testing in pairwise cross inventory, it was shown that the accuracy of the assessment of total shoot blight symptoms was fairly good in 2001–2002, with a κ statistic of 0.59–0.61. Severely damaged trees were identified with an agreement of κ = 0.81. The total area of pine forest both slightly and severely affected by G. abietina during 2001–2003 was estimated to be 484 000 ha. Three geographically separate damage centres were distinguished. Thus, despite a relatively sparse sample plot density, the national forest inventories have good potential for estimating the geographical distribution, areal extent, and dispersal of extensive damage outbreak. Results are dependent on the inventories being carried out with an accurate identification of target objects.

Interface of Tele-Task Operation for Automated Cultivation of Watermelon in Greenhouse

컴퓨터 시각 기술은 다양한 농작업 생력화에 있어 핵심적인 역할을 해왔다. 비록 컴퓨터 시각 기술이 광범위한 분야에 성공적으로 적용되고 있다고는 하지만 인간의 시각을 통한 인지 능력에 비하면 현재의 컴퓨터 시각 기술은 여전히 매우 미흡한 수준에 있다고 하겠다. 특히, 작업환경이 비구조적이고 가변적인 농작업 환경 하에서의 작업의 생력화는 이러한 기술적 문제를 극복하는 것이 작업의 성패를 좌우하게 된다. 본 논문에서는 원격작업 개념을 도입하여 작업자와 작업기계간의 호환적인 인터페이스를 구축하고 컴퓨터와 인간의 혼합형 의사결정 시스템을 구현하여 기존의 컴퓨터 시각 기술이 갖는 인지 처리 능력의 한계를 극복하는 시스템을 제안하였다. 시설재배에 요구되는 전정, 관수, 방제, 제초, 수확, 운반 등과 같은 다양한 작업들은 작업 대상체에 대한 인식을 바탕으로 수행된다. 특히 가변적인 자연 조명 환경 하에서 수박과 줄기 그리고 잎이 혼재되어 있는 재배현장의 영상으로부터 수박을 추출하여 그 위치 좌표를 산출하는 작업은 기술적으로 매우 어려운 작업이며 수박이 잎과 줄기로 덮혀 있는 경우 더욱 어려워진다. 제안한 개념을 구현하기 위하여 무선으로 수신되는 재배 현장의 수박 영상으로부터 수박을 인식하도록 하였다. 개발한 시스템은 작업자(농민), 컴퓨터 그리고 자동화 작업설비가 상대적으로 수월성을 갖는 기능을 중심으로 역할을 분담하도록 구축하였다 개발 시스템은 크게 무선원격 모니터링 및 작업제어 모듈, 무선원격 영상 획득 및 데이터 송수신 모듈, 작업자와 컴퓨터간의 인터페이스 모듈로 구성하였다 작업자는 RF 송수신 모듈을 통하여 무선으로 획득되어 터치 스크린에 보여지는 영상을 통하여 작업 지시를 하게 되고 이 작업 지시로부터 컴퓨터는 국부 영상처리 시퀀스를 통하여 수박을 추출하고 위치를 산출하게 된다. 개발한 인터페이스 시스템은 가변적이고 복잡한 작업 환경하에서 작업 대상체의 정보를 실시간으로 성공적으로 추출하였다. 제안한 원격작업 인터페이스 시스템은 다양한 생물생산 작업의 생력화를 촉진하는 중심적 역할을 할 것으로 기대된다. Computer vision technology has been utilized as one of the most powerful tools to automate various agricultural operations. Though it has demonstrated successful results in various applications, the current status of technology is still for behind the human's capability typically for the unstructured and variable task environment. In this paper, a man-machine interactive hybrid decision-making system which utilized a concept of tole-operation was proposed to overcome limitations of computer image processing and cognitive capability. Tasks of greenhouse watermelon cultivation such as pruning, watering, pesticide application, and harvest require identification of target object. Identifying water-melons including position data from the field image is very difficult because of the ambiguity among stems, leaves, shades. and fruits, especially when watermelon is covered partly by leaves or stems. Watermelon identification from the cultivation field image transmitted by wireless was selected to realize the proposed concept. The system was designed such that operator(farmer), computer, and machinery share their roles utilizing their maximum merits to accomplish given tasks successfully. And the developed system was composed of the image monitoring and task control module, wireless remote image acquisition and data transmission module, and man-machine interface module. Once task was selected from the task control and monitoring module, the analog signal of the color image of the field was captured and transmitted to the host computer using R.F. module by wireless. Operator communicated with computer through touch screen interface. And then a sequence of algorithms to identify the location and size of the watermelon was performed based on the local image processing. And the system showed practical and feasible way of automation for the volatile bio-production process.

Transient phase-locking of 40 Hz electrical oscillations in prefrontal and parietal human cortex reflects the process of conscious somatic perception

Electrical potential oscillations in the range of 35–45 Hz (gamma waves) have recently been shown to occur rather ubiquitously in the brain of awake humans. During selective somatic attention, we demonstrate a transient phase-locking of the gamma waves generated in the contralateral prefrontal and parietal cortical areas that we had previously shown to be involved in such selective attention tasks. In line with other microphysiological evidence obtained on mammalian visual cortex, this selective functional synchronization between critical human brain areas (as far as about 9 cm apart) is proposed to reflect the transient ‘binding’ of discrete cognitive features that are processed in distributed neuronal assemblies of the brain whereby the conscious perception of an object or event can be achieved. On this basis we emphasize that the conscious function of the brain is neither epiphenomenal nor delayed, but operates transiently to integrate relevant perceptual features at the time of target object identification and of conscious behavioural decision.