Content uploaded by James E. Cutting
Author content
All content in this area was uploaded by James E. Cutting
Content may be subject to copyright.
Recognizing the gender of walkers from point-lights mounted on ankles: Some second thoughts
... Studies in human perception have displayed walking motion using only dots of light located at the joints and have found test subjects quite adept at assessing the nature of the underlying motion [13]. In particular, subjects could identify the gender of a walker and recognize specific individuals even when no other cues were available [6,16,17]. ...
... They also reported that the gender of unfamiliar walkers was readily identifiable, even after the number of lights had been reduced to just two located on the ankles [16]. In a published note, they later explained that the two light-dot decisions were probably attributable to stride length [17]. Continuing this work, Barclay, Cutting, and Kozlowski [2] showed that gender recognition based on walking gaits required between 1.6 and 2.7 seconds of display, or about two step cycles. ...
Human figures have been animated using a wide variety of geometric models including stick figures, polygonal models, and NURBS-based models with muscles, flexible skin, or clothing. This paper reports on experiments designed to ascertain whether a viewer's perception of motion characteristics is affected by the geometric model used for rendering. Subjects were shown a series of paired motion sequences and asked if the two motions in each pair were "the same" or "different." The two motion sequences in each pair used the same geometric model. For each trial, the pairs of motion sequences were grouped into two sets where one set was rendered with a stick figure model and the other set was rendered with a polygonal model. Sensitivity measures for each trial indicate that for these sequences subjects were better able to discriminate motion variations with the polygonal model than with the stick figure model.
... Since biological studies [42] have demonstrated the effect of SHR on gender classification, we take SB =| j 12 − j 5 | and SH R = |j 12 −j 5 | |j 22 −j 18 | into account. Finally, we assign these anthropometric features to form a pseudo skeleton. ...
Gait recognition aims to identify people by the way they walk. Currently available gait recognition datasets mainly contain single-person gait data in relatively simple walking conditions, which limits research of robust gait recognition methods. In this paper, OG RGB+D dataset is presented to cope with this crucial limitation of other gait datasets. It includes the common walking conditions under occlusion in daily life, that is, those daily walking conditions in which people’s normal walking patterns are occluded, including self-occlusion caused by views, occlusion caused by clothing or objects, and mutual occlusion between people. The dataset provides multi-modal data to support different types of methods, collected by multiple Azure Kinect DK sensors using synchronous data acquisition system (Multi-Kinect SDAS). Moreover, we propose a model-based gait recognition method SkeletonGait for gait recognition in walking conditions under occlusion, which learns discriminative gait features from human dual skeleton model composed of skeleton and anthropometric features through a siamese Spatio-Temporal Graph Convolutional Network (siamese ST-GCN). The experimental results show that SkeletonGait surpasses state-of-the-art methods in the case of severe occlusion. We believe that the introduction of our dataset will enable the community to apply, adapt, and develop various robust gait recognition methods. The dataset will be available at https://github.com/cvNXE/OG-RGB-D-gait-dataset.
... Sur la base de ces expériences, les observateurs peuvent reconnaître différents types de mouvements humains tels que : marcher, sauter, danser et ainsi de suite. De plus, l'observateur peut porter un jugement sur le sexe de l'acteur [113], et même identifier la personne si elle connaît déjà sa démarche [74]. Cutting [45] a soutenu que la reconnaissance des mouvements est purement basée sur des caractéristiques dynamiques de la marche par opposition à des études antérieures qui ont été confondus par des indices de familiarité, la taille, la forme ou d'autres sources d'information. ...
The main focus of this research thesis is the automated recognition and analysis
of human activities from video sequences in order to determine what human
actions occur. It has recently emerged as a fundamental research topic in the field
of computer vision and machine learning. This problem is particularly difficult
due to the huge variations in the appearance and motion variations when performing
actions in addition to challenging factors related to the acquisition settings
as viewpoint, background clutters and occlusions. This is further exacerbated by
the huge amount of video data to analyse. Many applications are in urgent need
for vision-based solution to aid in the automated understanding of human actions,
including security, sports and even for autonomous cars.
In this thesis, we propose a motion descriptor based on optical flux estimation
for human action recognition, taking into account only the characteristics
derived from motion. The signature of human action consists of a histogram containing
kinematic features that include local and global traits. Experimental results
from the Weizmann and UCF101 datasets confirmed the potential of the proposed
approach with classification rates of 98.76% and 70% respectively to distinguish
between different human actions. For comparative and performance analysis, different
types of classifiers including Knn, decision tree and SVM are applied to the
proposed descriptors and deep Learning was also used. Further analysis is performed
to assess the proposed descriptors under different resolutions and frame
rates. The obtained results are in alignment with the early psychological studies
reporting that human motion is adequate for the perception of human activities.
... Manuscript submitted to ACM Since biological studies [18] have demonstrated the effect of SHR on gender classification , we take = |j 12 − j 5 | and = |j 12 −j 5 | |j 22 −j 18 | into account. Then we assign these anthropometric features to form a pseudo skeleton. ...
Human gait is one of important biometric characteristics for human identification at a distance. In practice, occlusion usually occurs and seriously affects accuracy of gait recognition. However, there is no available database to support in-depth research of this problem, and state-of-arts gait recognition methods have not paid enough attention to it, thus this paper focuses on gait recognition under occlusion. We collect a new gait recognition database called OG RGB+D database, which breaks through the limitation of other gait databases and includes multimodal gait data of various occlusions (self-occlusion, active occlusion, and passive occlusion) by our multiple synchronous Azure Kinect DK sensors data acquisition system (multi-Kinect SDAS) that can be also applied in security situations. Because Azure Kinect DK can simultaneously collect multimodal data to support different types of gait recognition algorithms, especially enables us to effectively obtain camera-centric multi-person 3D poses, and multi-view is better to deal with occlusion than single-view. In particular, the OG RGB+D database provides accurate silhouettes and the optimized human 3D joints data (OJ) by fusing data collected by multi-Kinects which are more accurate in human pose representation under occlusion. We also use the OJ data to train an advanced 3D multi-person pose estimation model to improve its accuracy of pose estimation under occlusion for universality. Besides, as human pose is less sensitive to occlusion than human appearance, we propose a novel gait recognition method SkeletonGait based on human dual skeleton model using a framework of siamese spatio-temporal graph convolutional networks (siamese ST-GCN). The evaluation results demonstrate that SkeletonGait has competitive performance compared with state-of-art gait recognition methods on OG RGB+D database and popular CAISA-B database.
Extensive research in social perception and biological motion has converged on the finding that humans are particularly accurate in identifying gender from the gait of minimal visual conspecific stimuli (e.g., point-light walkers). Despite the preponderance of evidence in favor of this ability, we return to the original paradigm and vary a single parameter—weight. Across nine pre-registered studies, participants (N = 3,196) were assigned to view the gait of point-light walkers based on actual human motion patterns. We find a decline in the accuracy of identifying the gender of female point-light walkers as their weight increases. However, as the weight of female walkers decreases, gender identification accuracy is recovered. These findings carry implications for the gendered nature of weight bias and the role of weight in human perception.
Previous work has shown that a number of factors can affect perceived attractiveness of opposite-sex dancers. For women watching men, body symmetry, perceived strength, vigor, skillfulness, and agility of movement, as well as greater variability and amplitude of the neck and trunk, are positively related to perceived attractiveness. For men watching women, body symmetry is also important, and femininity/masculinity of movement likely also plays a role for both sexes. Our aim here was to directly compare characteristics of attractive opposite-sex dancers under the same conditions. Sixty-two heterosexual adult participants (mean age = 24.68 years, 34 females) were presented with 48 short (30 s) audiovisual point-light animations of adults dancing to music. Stimuli were comprised of eight females and eight males, each dancing to three songs representative of Techno, Pop, and Latin genres. For each stimulus, participants rated perceived femininity/ masculinity as appropriate, sensuality, sexiness, mood, and interestingness of the dancer. Seven kinematic and kinetic features-downforce, hip wiggle, shoulder vs. hip angle, hip-knee phase, shoulder-hip ratio, hip-body ratio, and body symmetry-were computationally extracted from the stimuli. Results indicated that, for men watching women, hip-knee phase angle was positively related to ratings of perceived interestingness and mood, and hip-body ratio was positively related to ratings of perceived sensuality. For women watching men, downforce was positively related to ratings of perceived sensuality. Our results partially support previous work, and highlight some similarities and differences between male and female perceptions of attractiveness of opposite-sex dancers.
BACKGROUND: Generating locomotion for characters is a complex field with many challenges remaining for researchers to tackle. Whilst there has been various research undertaken into how to create diverse motion using physical simulation, inverse kinematics and motion capture, there is still little research on how to relate changes in virtual characters' body shape to the way they walk. This is important as audiences are capable of detecting repetition of character appearance and walking styles. By relating generated walk cycles to the body morphology of characters we can improve their believability. And to achieve this using a dynamic and automated system would save animators time when needing to create a variety of believable characters.
OBJECTIVEVS: This study will explore how people perceive gait to change over variations in body shape, how gait actually varies and whether it is possible to build a framework that believably correlates changes in gait parameters over changes in anthropometric parameters. Implementing this framework could then produce a tool for animators that generates a variety of virtual characters with believable variations in walking styles. The goal of this project is to improve the believability of virtual characters by relating virtual character's body shapes to an appropriate walk cycle.
METHODS: 8 papers were analysed to generate 8 empirical appearance to motion trendline formulas. These formulas formed the basis of the scripted animation tool. The animation tool was then used to create a point light survey testing 6 motion parameters to test people's perception of changes in motion over appearance. n= 59 participants completed the perceptual online video surveys. The animation tool's formulas were updated with the results of the point light survey and another survey was created using character meshes. n= 69 participants completed the perceptual online video surveys. 28 adult male gait patterns were motion captured and analysed using a Vicon motion capture suite. 3 5 parameters were analysed to have the strongest appearance to motion correlation and were sorted by order of perceptual dominance. These parameters were implemented in the scripted animation tool and a final perceptual poll was conducted. n= 96 participants validated the final animation tool using an online video survey.
FINDINGS: The empirical data analysis identified speed, stride length, step width, stance/ step phase and foot progression as motion parameters that change over increases in Body Mass Index (BMI). The point light perceptual survey found that changes to arm abduction, average arm bob and arm swing all produced motions associated with obese body morphologies. The character mesh perceptual survey verified that speed and walking base were motion parameters associated with changes in body morphology, whilst verifying previous parameter strengths and combinations. The actual motion capture sessions produced a framework of 5 appearance to motion formulas, ordered by perceptual dominance. The predictive correlations include: 1. Preferred walking speed over height 2. Average arm abduction over chest circumference 3. Walking base over waist-to-height ratio 4. Arm bob magnified over height 5. Arm swing over body fat percentage A final perceptual video poll found that when asked to rank 4 different types of obese generated motion, participants voted the framework of anthropometric to locomotive parameters tool to be the most believable by a 38% majority.
CONCLUSIONS: This study identifies 5 gait parameters that people have identified as being perceptually dominant. The motion capture analysis highlighted 5 gait parameters with significant correlations to appearance parameters. When implementing the chosen combination of appearance to gait parameters a significant majority of people ranked this to more believably represent an obese character walk, than a lean, obese and keyframe obese walk. An efficient and believable method for generating diverse locomotion that relates to the body morphology of the character has been created and validated.
Body feature, body movement such as gait characteristics, and anthropometric measures are commonly utilised to identify perpetrators. In the past, CCTV was the main device suited to obtain these, important particulars of perpetrators. However, advances in electronic pressure sensors in medical field, podiatry, physiotherapy and sport sciences have resulted in the availability of these sensors for crime surveillance applications. A research was carried out to develop a novel surveillance method of combining CCTV cameras and under-carpet floor pressure sensors to obtain certain body movement characteristics of crime perpetrator based on force/pressure distribution under the shoes. Thirteen adults were recruited and each subject was instructed to perform ten trials to obtain the needed foot force under normal shoes and foot pressure distance measures. To evaluate these novel measures' confidence, the mean within group and between group significant scores (P) were determined as: (1) 1.0; and (2) 0.001 respectively. These initial results showed that the novel method could be suitable for identifying perpetrators successfully.
The sex of human walkers can be recognized without familiarity cues from displays of pointlight sources mounted on major joints.
Static versions of these abstract displays do not permit accurate recognition of sex. Variation in the degree of armswing
or in walking speed generally interferes with recognition, except that faster speeds are associated somewhat with improved
recognition of females. Lights on upper-body joints permit more accurate guesses than do Lights on lower-body joints, but
identification is possible even from minimal displays, with lights placed only on the ankles. No feedback was given to observers.
Confidence judgments of sex relate to the accuracy of responses in a manner that suggests that viewers know what they are
doing.
Several temporal and spatial factors affect gender recognition of a walker when portrayed, without familiarity cues, as a
dynamic point-light display. We demonstrate that, among temporal parameters, the duration of the dynamic stimulus must be
longer than 1.6 sec, but that 2.7 sec is fully adequate. Given the speed of our walkers, the recognition threshold appears
to be roughly two step cycles. In addition, presentation rate of the stimulus must be near to normal, perhaps because nonnormal
rates alter apparent gravity and obscure the normal relationship between output and conservation of energy. We demonstrate
that, among spatial factors, the discreteness of the joint information must be maintained for accurate recognition. We go
on to argue that it is the information about the shoulder and the hip of a walker that is of primary importance. Finally,
inversion of the stimulus display produces the unexpected effect of reversing the apparent sex of most walkers. That is, when
presented upside down, male walkers appear female and female walkers appear male.
Viewers can determine the gender of a walker from sagitally projected, dynamic displays of point-lights attached to prominent joints. This article explores three interrelated approaches in search of a biomechanical invariant that viewers might use. The first, an index of torso structure, accounts for the data handsomely but seems inappropriate because it is not directly revealed in the dynamic stimuli. The second, a dynamic index of visible torsion in the trunk of a walker, also fits the data well but seems to have a logical problem and a difficulty in accounting for performance in certain conditions of several previous studies. The third has the strengths of the first two indices, and it can account for some other data as well. It is the center of moment and is a "deeper", more general description of the invariant. This center is a point around which all movement occurs. It can be thought of as one specification of the gestalt law of common fate and may be helpful for the study of movement perception in general.
Synthetic versions of human walkers were generated by computer as point-light displays. Previously it had been determined that the natural gaits of males and females differ according to the extent of movement at the shoulder and the hip. These movements were measured and then used to synthesize the stimuli used in the present study. These stimuli are shown here to be identified by untrained viewers as male when the shoulder movement is greater than the hip movement, and female when the configuration is reversed. Because of the coherence of the display lights representing the shoulder and hip are not necessary for gender recognition, although they do increase performance level. Hypernormality and heavy-footedness in gait are also discussed. Finally, all results are linked to an underlying biomechanical invariant, the center of moment.
Viewers can determine the gender of a walker from sagittally projected, dynamic displays of point-lights attached to prominent joints. This article explores three interrelated approaches in search of a biomechanical invariant that viewers might use. The first, an index of torso structure, accounts for the data handsomely but seems inappropriate because it is not directly revealed in the dynamic stimuli. The second, a dynamic index of visible torsion in the trunk of a walker, also fits the data well but seems to have a logical problem and a difficulty in accounting for performance in certain conditions of several previous studies. The third has the strengths of the first two indices, and it can account for some other data as well. It is the center of moment and is a "deeper," more general description of the invariant. This center is a point around which all movement occurs. It can be thought of as one specification of the gestalt law of common fate and may be helpful for the study of movement perception in general.
Generation of synthetic male and female walkers through manipulation of a biomechnical invariant A biomechanical invariant for gait perception Recognizing the sex of a walker from a dynamic point-light display
- J E Cutiing
- D R Proffiti
- L T Kozlowski
CUTIING, J. E. Generation of synthetic male and female walkers through manipulation of a biomechnical invariant. Perception, 1978, in press. CUTIING, 1. E., PROFFITI', D. R., & KOZLOWSKI, L. T. A biomechanical invariant for gait perception. Journal of Experi-mental Psychology: Human Perception and Performance, 1978, 4, in press. KOZLOWSKI, L. T., & CUTI'ING, J. E. Recognizing the sex of a walker from a dynamic point-light display. Perception & Psycho-physics, 1977, 21, 575-580. (Received for publication February 22, 1978; accepted February 23, 1978.)