基于目标检测的东北虎个体自动识别

东北虎个体的自动识别是种群数量评估和制定有效保护策略的重要基础。以东北虎林园和怪坡虎园38 只虎为研究对象,将目标检测方法首次应用到东北虎个体识别研究中,采用多种深度卷积神经网络模型,以实现虎个体的自动识别。首先通过相机在不同角度对 38 只东北虎进行拍摄取样,建立包含13579张图像的虎样本数据集。由于虎的体侧条纹信息不具有对称性,所以运用单次多盒目标检测（Single Shot MultiBox Detector, SSD）方法,对虎的躯干左侧条纹、右侧条纹以及脸部等不同部位图像,进行自动检测并分割提取,极大节省手工截取时间。在检测分割出的左右侧及脸部不同部位图片基础上,运用上、下、左、右平移变换进行数据增强,使图片数目扩大为原来的5 倍。采用LeNet、AlexNet、ZFNet、VGG16、ResNet34共5 种卷积神经网络模型进行个体自动识别。为了提高识别准确率,运用平均值和最大值不同组合方式来优化池化操作,并在全连接层引入概率分别为0.1、0.2、0.3、0.4的丢弃（Dropout）操作防止过拟合。实验表明,目标检测模型耗时较少,截取分割老虎不同部位条纹能达到0.6 s/张,远快于人工截取速度,并且在测试集上准确率能达到97.4%。不同姿态下的目标部位都能正确识别并分割。ResNet34模型的准确率优于其他网络模型,左右侧条纹以及脸部图像识别准确率分别为93.75%、97.01%和 86.28%,右侧条纹识别准确率优于左侧条纹和脸部图像。研究为野生虎自动相机影像的识别提供技术参考。在未来研究中,对东北虎个体影响数据进行扩充,选取更多影像数据进行训练,使网络具有更强的适应性,从而实现更准确的个体识别。     

Features effective in course control during object fixation by walking Colorado beetles

A test paradigm is designed to determine which features of viewed objects (a disk 10° in diameter) are responsible for preferential fixation by Colorado beetles (Leptinotarsa decemlineata). The beetle walks on a locomotion compensator that permits free choice of direction in a visual surround containing three different objects. The beetles use both colour and contrast information in these choice experiments. The preferred colours are yellow and orange; their attractiveness is considerably enhanced when they appear in a pattern with highly contrasting features such as stripes. In its optimal form, the object amounts to a dummy beetle, with a pattern that resembles the markings on the elytra of this species. Since the beetles appear to employ a contact pheromone for sex recognition, the function of orientation to these patterns is unclear. However, the patterns might accelerate the identification of conspecifics or be effective at a distance. The reaction to the dummies is telotactic — that is, the beetles walk towards the dummy. One other object, a 10°-wide black stripe, elicits menotactic orientation, in which the beetle maintains a stable course at an angle to the object.     

Spatial aspects of spontaneous object grouping by young chimpanzees(Pan troglodytes)

I report spontaneous spatial object grouping in five chimpanzees (Pan troglodytes)ranging from 1 to 4 years of age. I recorded subjects' spontaneous interactions with groups of objects, noting the spatial relations in the groupings and the constructive processes that the subjects adopted. Though one subject developed vertical stacks, none of them realized horizontal alignments or spatial correspondences between groups. All subjects showed consistent trends in their constructive processes toward manipulating objects in relation to the body: they increasingly manipulated objects globally, that is to say, they moved or held objects together with the same part of the body. They also increasingly placed objects on the body or in correspondence to their symmetrical body parts. The increasing importance assigned by chimpanzees to body- object relations over object- object relations contrasts with human developmental trends in the domain of manipulative space.     

Contrasting responses to object orientation and illumination by the blowflies Chrysomya chloropyga and Lucilia sericata

The alighting response of two species of blowfly, Chrysomya chloropyga (Weidemann) and Lucilia sericata(Meigen) (Diptera: Calliphoridae), were considered in relation to the orientation and illumination of a flat, oblong visual object. A strong response to orientation was displayed by L. sericata with the greatest number alighting on the vertical object; there was no significant effect of orientation on the number of C. chloropyga which alighted. Both species alighted preferentially on the more strongly illuminated side of a vertical object. With the object horizontal and more strongly illuminated from below, C. chloropygaagain landed on the most brightly illuminated side. In contrast, reducing the brightness on the upper surface of the horizontally-suspended object reduced the number of L. sericataalighting on the upper surface but also reduced the overall catch, suggesting that for L. sericata the response to the illumination of a surface does not outweigh its response to its orientation. The implications of these results for the development of trapping technology for the monitoring or control of these economically important pest species is discussed.     

Social object play among young Japanese macaques (Macaca fuscata) in Arashiyama, Japan

Social object play (SOP), i.e., social play using portable object(s), among young Japanese macaques (Macaca fuscata; 0-4 years old) in the Arashiyama E troop was studied using a modified sequence sampling method from July to October 2000. SOP was a relatively common activity for most of the young macaques and often continued for long periods. Participants used many kinds of object, including edible natural objects and artificial objects, such as plastic bottles, but they never used provisioned food or wild fruit in SOP bouts. An analysis of long bouts (>/=0.5 min) revealed the following interactive SOP features: (1) at any given time, participants used only one object, and only one participant held the object; (2) during SOP play-chasing, the object holder was likely to be chased by others; (3) during long bouts, the object changed hands frequently; and (4) agonistic competition for an object among young macaques was rare. Combinations of sexes, ages, relative ranks, or matrilines of the object holder and non-holder did not affect the tendency that the holder was chased by non-holder(s) during play-chasing. Even when there was a change in object holders, the repetitiveness of this interactive pattern, i.e., that the holder would be chased during SOP bouts, distinguished the SOP structure from that of other types of social play without object(s). General proximate social play mechanisms, such as self-handicapping or role taking, were associated with SOP. Other mechanisms that affected SOP included the following: (1) young macaques treated an object as a target in play competition, and (2) 'being the holder of a target object' was associated with the 'role of the chasee.'     

Further report on object manipulation in non-human primates: A comparison within 13 species of the genusMacaca

Thirteen captive species of the genusMacaca were presented with certain non-edible objects and the subsequent manipulation was observed. The following results were obtained: (1) Thefascicularis group species displayed a great variety in rolling the manipulandum, which has been reported to be one of the actions characterizing cercopithecine species; (2) manipulating the object with one finger was typical of thesilenus-sylvanus andsinica groups, whereas use of the fingers moving opposably was common to all species ofMacaca; (3) secondary manipulations in relation to the wire-mesh were observed only inM. silenus, M. sinica, andM. nemestrina, whereas those in relation to water were common to allMacaca species; and (4) within-subgroup differences were great in thesilenus-sylvanus andsinica groups as compared to thefascicularis group.     

Active electroreception in Gymnotus omari: Imaging, object discrimination, and early processing of actively generated signals

Weakly electric fishes “electrically illuminate” the environment in two forms: pulse fishes emit a succession of discrete electric discharges while wave fishes emit a continuous wave. These strategies are present in both taxonomic groups of weakly electric fishes, mormyrids and gymnotids. As a consequence one can distinguish four major types of active electrosensory strategies evolving in parallel. Pulse gymnotids have an electrolocating strategy common with pulse mormyrids, but brains of pulse and wave gymnotids are alike. The beating strategy associated to other differences in the electrogenic system and electrosensory responses suggests that similar hardware might work in a different mode for processing actively generated electrosensory images. In this review we summarize our findings in pulse gymnotids’ active electroreception and outline a primary agenda for the next research.     

Effects of early experience on the behaviour of yearling rhesus monkeys (Macaca mulatta) in the presence of a strange object: classification and correlation approaches

Special techniques are needed to help us understand the long-term effects of infants' experiences of their relationships with their mothers, especially when direct experimental control of the interaction between mother and young is impossible, or could distort their relationship. A classification approach is developed to show how outcomes in individuals can be predicted from their earlier experiences and characteristics. Four-week-old rhesus monkey infants' characteristic levels of enterprise persisted through their first year, but could be reduced in certain individuals who had been kept off their mothers by high rates of early maternal rejection. Correlational approaches are difficult to interpret when they fail to confirm simple effects of experience or of individual characteristics, because they are based on groups not individuals, and because certain individuals can affect the values of correlation coefficients in ways that cannot be specified without using a classification approach. The possible effects on correlation coefficients of inconsistent individuals, and of being unable to specify which infants respond in which ways to relevant variables were discussed.     

Electrophysiological correlates of object location and object identity processing in spatial scenes

The ability to quickly detect changes in our surroundings has been crucial to human adaption and survival. In everyday life we often need to identify whether an object is new and if an object has changed its location. In the current event-related potential (ERP) study we investigated the electrophysiological correlates and the time course in detecting different types of changes of an objecṫs location and identity. In a delayed match-to-sample task participants had to indicate whether two consecutive scenes containing a road, a house, and two objects, were either the same or different. In six randomly intermixed conditions the second scene was identical, one of the objects had changed its identity, one of the objects had changed its location, or the objects had switched locations. The results reveal different time courses for the processing of identity and location changes in spatial scenes. Whereas location changes elicited a posterior N2 effect, indicating early mismatch detection, followed by a P3 effect reflecting post-perceptual processing, identity changes elicited an anterior N3 effect, which was delayed and functionally distinct from the N2 effect found for the location changes. The condition in which two objects switched position elicited a late ERP effect, reflected by a P3 effect similar to that obtained for the location changes. In sum, this study is the first to cohesively show different time courses for the processing of location changes, identity changes, and object switches in spatial scenes, which manifest themselves in different electrophysiological correlates.     

Low-level correlations between object properties and viewpoint can cause viewpoint-dependent object recognition

Viewpoint-dependent recognition performance of 3-D objects has often been taken as an indication of a viewpoint-dependent object representation. This viewpoint dependence is most often found using metrically manipulated objects. We aim to investigate whether instead these results can be explained by viewpoint and object property (e.g. curvature) information not being processed independently at a lower level, prior to object recognition itself. Multidimensional signal detection theory offers a useful framework, allowing us to model this as a low-level correlation between the internal noise distributions of viewpoint and object property dimensions. In Experiment 1, we measured these correlations using both Yes/No and adjustment tasks. We found a good correspondence across tasks, but large individual differences. In Experiment 2, we compared these results to the viewpoint dependence of object recognition through a Yes/No categorization task. We found that viewpoint-independent object recognition could not be fully reached using our stimuli, and that the pattern of viewpoint dependence was strongly correlated with the low-level correlations we measured earlier. In part, however, the viewpoint was abstracted despite these correlations. We conclude that low-level correlations do exist prior to object recognition, and can offer an explanation for some viewpoint effects on the discrimination of metrically manipulated 3-D objects.     

Analysis of object clustering

Objects of a given type lie on a plane or in a thin layer. In the analysis of the arrangement of such objects, the question may arise as to whether these objects form clusters or whether they are randomly distributed. This question can be answered by the calculation of the coefficient of dispersion (CD) of the objects. Some statistical properties of this coefficient, i.e., its sensitivity and its chi-square approximation, are analyzed in this paper. The proposed approach for calculating sensitivity is a mathematical simulation that is not generally applicable but is valid in various biologic applications. Assuming that the objects are much smaller than the expected clusters and that no knowledge exists about interactions among the objects or between the objects and their surroundings, the objects can be assumed to be distributed randomly on the plane. Such situations may occur in various fields, i.e., biology, sociology, archaeology and ecology. The authors have applied this approach successfully in investigating the densities of T-cell lymphocytes in renal transplants treated with Cyclosporin A or conventional immunosuppressive therapy.     

Disruptive camouflage impairs object recognition

Whether hiding from predators, or avoiding battlefield casualties, camouflage is widely employed to prevent detection. Disruptive coloration is a seemingly well-known camouflage mechanism proposed to function by breaking up an object''s salient features (for example their characteristic outline), rendering objects more difficult to recognize. However, while a wide range of animals are thought to evade detection using disruptive patterns, there is no direct experimental evidence that disruptive coloration impairs recognition. Using humans searching for computer-generated moth targets, we demonstrate that the number of edge-intersecting patches on a target reduces the likelihood of it being detected, even at the expense of reduced background matching. Crucially, eye-tracking data show that targets with more edge-intersecting patches were looked at for longer periods prior to attack, and passed-over more frequently during search tasks. We therefore show directly that edge patches enhance survivorship by impairing recognition, confirming that disruptive coloration is a distinct camouflage strategy, not simply an artefact of background matching.     

Neural mechanisms of object recognition

Single-unit recordings from behaving monkeys and human functional magnetic resonance imaging studies have continued to provide a host of experimental data on the properties and mechanisms of object recognition in cortex. Recent advances in object recognition, spanning issues regarding invariance, selectivity, representation and levels of recognition have allowed us to propose a putative model of object recognition in cortex.     

Bayesian models of object perception

The human visual system is the most complex pattern recognition device known. In ways that are yet to be fully understood, the visual cortex arrives at a simple and unambiguous interpretation of data from the retinal image that is useful for the decisions and actions of everyday life. Recent advances in Bayesian models of computer vision and in the measurement and modeling of natural image statistics are providing the tools to test and constrain theories of human object perception. In turn, these theories are having an impact on the interpretation of cortical function.     

Neuropsychological studies of object recognition

It is well established that disorders of visual perception are associated with lesions in the right hemisphere. Performances on tasks as disparate as the identification of objects from unusual views of objects drawn so as to overlap, of fragmented letters, of familiar faces, and of anomalous features in drawings, have been shown to be impaired in patients with focal right posterior lesions. A series of investigations are reviewed, directed towards analysing the basis of these deficits. Explanations in terms of primary visual impairment can be rejected, as can an account in terms of faulty figure-ground organization. It is argued that a wide variety of such perceptual deficits--all of which are concerned with meaningful visual stimuli--can be encompassed by the notion of faulty perceptual categorization at an early post-sensory stage of object recognition. Moreover, there is evidence suggesting that some of these various perceptual deficits can be mutually dissociated. The concept of perceptual categorization is discussed in the wider context of tentative model of object recognition.     

Models of object recognition.

Progress in the understanding of visual recognition in the past year has been signified by the demonstration of computational feasibility of and psychophysical support for two-dimensional view-interpolation methods.