Complex Visual Environments Research Articles

Visual SLAM in Long-Range Autonomous Parking Application Based on Instance-Aware Semantic Segmentation via Multi-Task Network Cascades and Metric Learning Scheme

<div class="section abstract"><div class="htmlview paragraph">Long-range Autonomous Parking is becoming an attractive application in terms of demands. The vehicle is capable of driving autonomously into the appointed parking slot when the driver leaves it at the drop-off spot. In this application, the ability of accurate localization has become a key issue, especially in GPS-denied environments. This paper proposes a method of localization and mapping for Long-range Autonomous Parking, which is achieved by Visual SLAM based on deep learning algorithms. Firstly, we propose an instance segmentation via multi-task network cascades, and even in a complex visual environment, the main roadway instances of interest in the parking lot IPM image can be detected, such as parking corners, speed bumps. Then we combine the information of wheel encoders to build a global semantic map of the parking lot. Vehicles can often rely on semantic map matching to achieve high-precision localization. However, without a good initial position, it is difficult to infer an accurate position by matching the semantic map, such as randomly selecting entrances to enter the parking lot. Therefore, we propose an area feature network based on metric learning to extract features that distinguish different areas and infer the approximate initial position of the vehicle. Specifically, we extract features from the images of the surround-view cameras, use the vehicle position as weak supervision, and finally construct an area feature map. In summary, our proposed method provides accurate vehicle localization and parking lot maps for Long-range Autonomous Parking.</div></div>

Phasic alerting increases visual processing speed in amnestic mild cognitive impairment

External warning cues temporarily increase the brain's sensitivity for upcoming events, helping individuals to flexibly adapt their reactions to the requirements of complex visual environments. Previous studies reported that younger and cognitively normal older adults profit from phasic alerting cues. Such an intact phasic alerting mechanism could be even more relevant in individuals with Alzheimer's disease who are characterized by reduced processing capacities. The present study employed a theory of visual attention based verbal whole report paradigm with auditory cues in order to investigate phasic alerting effects in amnestic mild cognitive impairment (aMCI). Patients with aMCI were also compared to a previously reported sample of cognitively normal older adults. In patients with aMCI, visual processing speed was higher in the cue compared to the no-cue condition. Further, visual processing speed was reduced in patients with aMCI compared to cognitively normal older adults. Taken together, the results suggest that the processing system of patients with aMCI exhibits general declines but can still integrate auditory warning signals on a perceptual level.

Retinal organization and visual abilities for flower foraging in swallowtail butterflies.

Papilio butterflies' ability to forage for flowers relies upon multiple visual cues such as color, brightness, and motion. Papilio learns the color of rewarding flowers and detects it at a distance. Its color vision is based on four photoreceptor classes: UV, blue, green, and red, providing sensitive wavelength discrimination. These four receptor classes also contribute to the perception of brightness and polarization. Papilio's motion vision is based on a different set of receptors: green, red, and broad band. This implies that two visual pathways exist in Papilio. The contribution of several receptor classes not only for chromatic vision but also achromatic vision likely enhances the butterfly's ability to detect flowers in complex visual environments.

Cervicothoracic Mechanical Impairment as Part of Complete Neurological Fall Risk Appraisal.

Assessment of individuals at risk for falling entails comprehensive neurological and vestibular examinations. Chronic limitation in cervical mobility reduces gaze accuracy, potentially impairing navigation through complex visual environments. Additionally, humans with scoliosis have altered otolithic vestibular responses, causing imbalance. We sought to determine whether dynamic cervical mobility restrictions or static cervicothoracic impairments are also fall risk factors. We examined 435 patients referred for soft-tissue musculoskeletal complaints; 376 met criteria for inclusion (mean age 52; 266 women). Patients were divided into nonfallers, single fallers, and multiple fallers, less or greater than 65 years old. Subject characteristics, dynamic cervical rotations, and static cervicothoracic axial measurements were compared between groups. Fear of falling was evaluated using the Falls Efficacy Scale-International questionnaire. Long-standing cervicothoracic pain and stiffness conferred increased risk of falling. Neck rotation amplitudes decreased with longer duration musculoskeletal symptoms and were significantly more restricted in fallers, doubling the risk of falling and contributing to increased fear of falling. Mid-thoracic scoliosis amplitudes increased over time, but static axial abnormalities were not greater among fallers, although thoracic kyphoscoliosis heightened fear of falling. In patients at fall risk, thoracic kyphoscoliosis and dynamic neck movements should be assessed, in addition to standard vestibular and neurological evaluations. Additionally, patients with soft tissue cervicothoracic pain and restricted mobility have increased fall frequency and fear of falling, independent of other fall risk factors and should undergo complete fall risk appraisal.

Visual Fixations and Motion Sensitivity: Protocol for an Exploratory Study.

BackgroundMotion sensitivity after vestibular disorders is associated with symptoms of nausea, dizziness, and imbalance in busy environments. Dizziness and imbalance are reported in places such as supermarkets and shopping malls which have unstable visual backgrounds; however, the mechanism of motion sensitivity is poorly understood.ObjectiveThe main aim of this exploratory observational study is to investigate visual fixations and postural sway in response to increasingly complex visual environments in healthy adults and adults with motion sensitivity.MethodsA total of 20 healthy adults and 20 adults with motion sensitivity will be recruited for this study. Visual fixations, postural sway, and body kinematics will be measured with a mobile eye tracker device, force plate, and 3D motion capture system, respectively. Participants will be exposed to experimental tasks requiring visual fixation on letters, projected on a range of backgrounds on a large screen during quiet stance. Descriptive statistics (mean and standard deviation) will be calculated for each of the variables. One-way independent-measures analyses of variance will be performed to investigate the differences between groups for all variables.ResultsData collection was started in May 2019 and was completed by February 2020. It was approved by Health and Disability Ethics Committees, Ministry of Health, New Zealand on November 2, 2018 (Ethics ref: 18/CEN/193). We are currently processing the data and will begin data analysis in July 2020. We expect the results to be available for publication by the end of 2020. The trial was funded by the Neurology Special Interest Group, Physiotherapy New Zealand, and the Eisdell Moore Centre in November 2018.ConclusionsThis study will provide a detailed investigation of visual fixations in response to increasingly complex visual environments. Investigating characteristics of visual fixations in healthy adults and those with motion sensitivity will provide insight into this disabling condition and may inform the development of new intervention strategies which explicitly cater to the needs of this population.Trial RegistrationAustralian New Zealand Clinical Trials Registry, ACTRN12619000254190; https://tinyurl.com/yxbn7nksInternational Registered Report Identifier (IRRID)PRR1-10.2196/16805

Parking Slot Detection on Around-View Images Using DCNN.

Due to the complex visual environment and incomplete display of parking slots on around-view images, vision-based parking slot detection is a major challenge. Previous studies in this field mostly use the existing models to solve the problem, the steps of which are cumbersome. In this paper, we propose a parking slot detection method that uses directional entrance line regression and classification based on a deep convolutional neural network (DCNN) to make it robust and simple. For parking slots with different shapes and observed from different angles, we represent the parking slot as a directional entrance line. Subsequently, we design a DCNN detector to simultaneously obtain the type, position, length, and direction of the entrance line. After that, the complete parking slot can be easily inferred using the detection results and prior geometric information. To verify our method, we conduct experiments on the public ps2.0 dataset and self-annotated parking slot dataset with 2,135 images. The results show that our method not only outperforms state-of-the-art competitors with a precision rate of 99.68% and a recall rate of 99.41% on the ps2.0 dataset but also performs a satisfying generalization on the self-annotated dataset. Moreover, it achieves a real-time detection speed of 13 ms per frame on Titan Xp. By converting the parking slot into a directional entrance line, the specially designed DCNN detector can quickly and effectively detect various types of parking slots.

Vacant Parking Slot Detection in the Around View Image Based on Deep Learning.

Due to the complex visual environment, such as lighting variations, shadows, and limitations of vision, the accuracy of vacant parking slot detection for the park assist system (PAS) with a standalone around view monitor (AVM) needs to be improved. To address this problem, we propose a vacant parking slot detection method based on deep learning, namely VPS-Net. VPS-Net converts the vacant parking slot detection into a two-step problem, including parking slot detection and occupancy classification. In the parking slot detection stage, we propose a parking slot detection method based on YOLOv3, which combines the classification of the parking slot with the localization of marking points so that various parking slots can be directly inferred using geometric cues. In the occupancy classification stage, we design a customized network whose size of convolution kernel and number of layers are adjusted according to the characteristics of the parking slot. Experiments show that VPS-Net can detect various vacant parking slots with a precision rate of 99.63% and a recall rate of 99.31% in the ps2.0 dataset, and has a satisfying generalizability in the PSV dataset. By introducing a multi-object detection network and a classification network, VPS-Net can detect various vacant parking slots robustly.

Does valence influence perceptual bias towards incongruence during binocular rivalry?

The efficient processing of complex visual environments is essential for effective functioning. In the natural environment, processing the context is as important as the processing of the target object since no object can be found in isolation. Congruent object-context associations in a visual scene facilitate object recognition, whereas incongruent associations decrease performance accuracy. Although there is a performance reduction, incongruent scenes are reported to have a perceptual bias due to the reallocation of the attentional resources towards the associated semantic conflict. Another key attribute that prepares the visual system to identify the important aspects of the environment is valence, and any visual scene can be classified into one of the three valence categories. Hence, the current study was designed to investigate how valence influences the perceptual bias towards incongruent object-context associations. An intermittent binocular rivalry task was used to measure the perceptual bias across valence categories. The results revealed a significant predominance of incongruent pictures when the associated valence was negative and neutral and remained unbiased for positive valence. We propose a valence-congruency interaction in which perceptual bias towards incongruence is greatly influenced by valence.

Analysis of Signage using Eye-Tracking Technology

Signs, in all their forms and manifestations, provide visual communication for wayfinding, commerce, and public dialogue and expression. Yet, how effectively a sign communicates and ultimately elicits a desired reaction begins with how well it attracts the visual attention of prospective viewers. This is especially the case for complex visual environments, both outside and inside of buildings. This paper presents the results of an exploratory research design to assess the use of eye-tracking (ET) technology to explore how placement and context affect the capture of visual attention. Specifically, this research explores the use of ET hardware and software in real-world contexts to analyze how visual attention is impacted by location and proximity to geometric edges, as well as elements of contrast, intensity against context, and facial features. Researchers also used data visualization and interpretation tools in augmented reality environments to anticipate human responses to alternative placement and design. Results show that ET methods, supported by the screen-based and wearable eye-tracking technologies, can provide results that are consistent with previous research of signage performance using static images in terms of cognitive load and legibility, and ET technologies offer an advanced dynamic tool for the design and placement of signage.

Automatic Recognition of Highway Tunnel Defects Based on an Improved U-Net Model

It is critically important for the safe operation of highway tunnels that defects in tunnel linings be promptly identified. Recently, many effective identification models based on deep learning technology have emerged in the field of tunnel defect recognition. However, the results of previous studies indicate that imperfections exist in these models in which the boundaries of the defects can only be generally located in the complex visual environment of tunnel linings, which include noisy patterns and uneven light sources. In order to more accurately identify defect details in such complex environments, an improved U-net model including a combined Squeeze-and-Excitation (SE) and ResNet block is proposed and evaluated in this paper. The accuracy of the conventional U-net model is found to be considerably improved by the inclusion of the SE and ResNet blocks. An under-sampling strategy is utilized to address the problem of imbalanced crack image data and determine an appropriate ratio of negative to positive samples for the crack dataset. The effectiveness of the proposed model is then demonstrated using actual highway tunnel lining images to compare its results with those of the Gabor Filter, Multiscale DCNN, FCN, and conventional U-net defect recognition models. The accuracy of the proposed model was found to be quite competitive with extant models.

How is Attention Deployed in a Complex Visual Environment?

How is Attention Deployed in a Complex Visual Environment?

Figure-Ground Modulation in the Human Lateral Geniculate Nucleus Is Distinguishable from Top-Down Attention

Nearly all of the information that reaches the primary visual cortex (V1) of the brain passes from the retina through the lateral geniculate nucleus (LGN) of the thalamus. Although the LGN's role in relaying feedforward signals from the retina to the cortex is well understood [1, 2], the functional role of the extensive feedback it receives from the cortex has remained elusive [3-6]. Here, we investigated whether corticothalamic feedback may contribute to perceptual processing in the LGN in a manner that is distinct from top-down effects of attention [7-10]. We used high-resolution fMRI at 7 Tesla to simultaneously measure responses to orientation-defined figures in the human LGN and V1. We found robust enhancement of perceptual figures throughout the early visual system, which could be distinguished from the effects of covert spatial attention [11-13]. In a secondexperiment, we demonstrated that figure enhancement occurred in the LGN even when the figure and surrounding background were presented dichoptically (i.e., to different eyes). As binocular integration primarily occurs in V1 [14,15], these results implicate a mechanism of automatic, contextually sensitive feedback from binocular visual cortex underlying figure-ground modulation in the LGN. Our findings elucidate the functional mechanisms of this core function of the visual system [16-18], which allows people to segment and detect meaningful figures in complex visual environments. The involvement of the LGN in this rich, contextually informed visual processing-despite showing minimal feedforward selectivity for visual features [19, 20]-underscores the role of recurrent processing at the earliest stages of visual processing.

Long-range neural inhibition and stimulus competition in the archerfish optic tectum.

The archerfish, which is unique in its ability to hunt insects above the water level by shooting a jet of water at its prey, operates in a complex visual environment. The fish needs to quickly select one object from among many others. In animals other than the archerfish, long-range inhibition is considered to drive selection. As a result of long-range inhibition, a potential target outside a neuron's receptive field suppresses the activity elicited by another potential target within the receptive field. We tested whether a similar mechanism operates in the archerfish by recording the activity of neurons in the optic tectum while presenting a target stimulus inside the receptive field and a competing stimulus outside the receptive field. We held the features of the target constant while varying the size, speed, and distance of the competing stimulus. We found cells that exhibit long-range inhibition; i.e., inhibition that extends to a significant part of the entire visual field of the animal. The competing stimulus depressed the firing rate. In some neurons, this effect was dependent on the features of the competing stimulus. These findings suggest that long-range inhibition may play a crucial role in the target selection process in the archerfish.

Rapid Extraction of Emotion Regularities from Complex Scenes in the Human Brain

Adaptive behavior requires the rapid extraction of behaviorally relevant information in the environment, with particular emphasis on emotional cues. However, the speed of emotional feature extraction from complex visual environments is largely undetermined. Here we use objective electrophysiological recordings in combination with frequency tagging to demonstrate that the extraction of emotional information from neutral, pleasant, or unpleasant naturalistic scenes can be completed at a presentation speed of 167 ms (i.e., 6 Hz) under high perceptual load. Emotional compared to neutral pictures evoked enhanced electrophysiological responses with distinct topographical activation patterns originating from different neural sources. Cortical facilitation in early visual cortex was also more pronounced for scenes with pleasant compared to unpleasant or neutral content, suggesting a positivity offset mechanism dominating under conditions of rapid scene processing. These results significantly advance our knowledge of complex scene processing in demonstrating rapid integrative content identification, particularly for emotional cues relevant for adaptive behavior in complex environments.

Faster eye movements in children with autism spectrum disorder.

Atypical visual exploration of both social and nonsocial scenes is often reported in Autism Spectrum Disorder (ASD) with less precise and longer saccades, potentially reflecting difficulties in oculomotor control. To assess a subset of oculomotor functions in ASD, 20 children with ASD and 21 age-matched typically developing (TD) children (2.6-11.5 years) partook in three tasks of increasing complexity, while no explicit instruction was provided: a prosaccade gap task, a color and a "categorical" visual search tasks (a face among butterflies and vice-versa). In addition to classical saccade metrics, we measured Distance error, (the distance between the target and the closest gaze position) and Time-to-target (the time taken to reach the target). In the prosaccade task, children with ASD were as accurate as TD children, yet faster to reach the stimulus. In the color visual search task, children with ASD were faster but less precise than TD children. In the categorical visual search, while TD children were more precise in orienting their gaze towards the face, children with ASD performed similarly in the two conditions; Time-to-target did not differ. Our results provide contradictory evidence regarding enhanced visual search ability in ASD: when considering response times, enhanced visual search performance was found in one task only, while when considering gaze precision no advantage was found. These three experiments demonstrate that the automatic saccadic system may function more rapidly in children with ASD. Nonetheless, a diminished sensitivity to bottom-up saliency and top-down influence might suppress this advantage in more complex visual environments. Autism Res 2019, 12: 212-224 © 2018 International Society for Autism Research, Wiley Periodicals, Inc. LAY SUMMARY: Three experiments with no instructions were designed to assess oculomotor functions in children with Autism Spectrum Disorder (ASD). In a saccade task, children with ASD were faster than but as accurate as control children. In visual search tasks, accuracy and speed decreased with increasing complexity of visual environment. Children with ASD showed faster automatic visual orientation, but this might hinder exploratory behaviors, leading to difficulties in complex and social situations.

Visual statistical learning at basic and subordinate category levels in real-world images.

Visual statistical learning (VSL) has been proposed as a powerful mechanism underlying the striking ability of human observers to handle complex visual environments. Previous studies have shown that VSL can occur when statistical information is embedded at multiple levels of abstraction, such as at semantically different category levels. In the present study, we further examined whether statistical regularities at a basic category level (e.g., a regular sequence of a bird, then a car, and then a dog) could influence the ability to extract statistical regularities at the subordinate level (e.g., a regular sequence of a parrot, then a sports car, and then an Eskimo dog). In the familiarization phase, participants were exposed to a stream of real-world images whose semantic categories had temporal regularities. Importantly, the temporal regularities existed at both the basic and subordinate levels, or the regularities existed at only the subordinate level, depending on the experimental condition. After completing the familiarization, participants performed a surprise two-alternative forced choice (2AFC) task for a familiarity judgment between two triplets in which the temporal regularities were either preserved or not preserved. Our results showed that the existence of statistical regularities at the basic level did not influence VSL at the subordinate level. The subsequent experiments showed these results consistently even when the basic-level categories had to be explicitly recognized and when the stimuli were not easily categorized at their subordinate level. Our results suggest that VSL is constrained to learn a particular level of patterns when patterns are presented across multiple levels.

Selective Attention to Faces in a Rapid Visual Stream: Hemispheric Differences in Enhancement and Suppression of Category-selective Neural Activity.

In daily life, efficient perceptual categorization of faces occurs in dynamic and highly complex visual environments. Yet the role of selective attention in guiding face categorization has predominantly been studied under sparse and static viewing conditions, with little focus on disentangling the impact of attentional enhancement and suppression. Here we show that attentional enhancement and suppression exert a differential impact on face categorization supported by the left and right hemispheres. We recorded 128-channel EEG while participants viewed a 6-Hz stream of object images (buildings, animals, objects, etc.) with a face image embedded as every fifth image (i.e., OOOOFOOOOFOOOOF…). We isolated face-selective activity by measuring the response at the face presentation frequency (i.e., 6 Hz/5 = 1.2 Hz) under three conditions: Attend Faces, in which participants monitored the sequence for instances of female faces; Attend Objects, in which they responded to instances of guitars; and Baseline, in which they performed an orthogonal task on the central fixation cross. During the orthogonal task, face-specific activity was predominantly centered over the right occipitotemporal region. Actively attending to faces enhanced face-selective activity much more evidently in the left hemisphere than in the right, whereas attending to objects suppressed the face-selective response in both hemispheres to a comparable extent. In addition, the time courses of attentional enhancement and suppression did not overlap. These results suggest the left and right hemispheres support face-selective processing in distinct ways-where the right hemisphere is mandatorily engaged by faces and the left hemisphere is more flexibly recruited to serve current tasks demands.

Feature reliability determines specificity and transfer of perceptual learning in orientation search.

Training can modify the visual system to produce a substantial improvement on perceptual tasks and therefore has applications for treating visual deficits. Visual perceptual learning (VPL) is often specific to the trained feature, which gives insight into processes underlying brain plasticity, but limits VPL’s effectiveness in rehabilitation. Under what circumstances VPL transfers to untrained stimuli is poorly understood. Here we report a qualitatively new phenomenon: intrinsic variation in the representation of features determines the transfer of VPL. Orientations around cardinal are represented more reliably than orientations around oblique in V1, which has been linked to behavioral consequences such as visual search asymmetries. We studied VPL for visual search of near-cardinal or oblique targets among distractors of the other orientation while controlling for other display and task attributes, including task precision, task difficulty, and stimulus exposure. Learning was the same in all training conditions; however, transfer depended on the orientation of the target, with full transfer of learning from near-cardinal to oblique targets but not the reverse. To evaluate the idea that representational reliability was the key difference between the orientations in determining VPL transfer, we created a model that combined orientation-dependent reliability, improvement of reliability with learning, and an optimal search strategy. Modeling suggested that not only search asymmetries but also the asymmetric transfer of VPL depended on preexisting differences between the reliability of near-cardinal and oblique representations. Transfer asymmetries in model behavior also depended on having different learning rates for targets and distractors, such that greater learning for low-reliability distractors facilitated transfer. These findings suggest that training on sensory features with intrinsically low reliability may maximize the generalizability of learning in complex visual environments.

Anatomical Analysis of the Retinal Specializations to a Crypto-Benthic, Micro-Predatory Lifestyle in the Mediterranean Triplefin Blenny Tripterygion delaisi.

The environment and lifestyle of a species are known to exert selective pressure on the visual system, often demonstrating a tight link between visual morphology and ecology. Many studies have predicted the visual requirements of a species by examining the anatomical features of the eye. However, among the vast number of studies on visual specializations in aquatic animals, only a few have focused on small benthic fishes that occupy a heterogeneous and spatially complex visual environment. This study investigates the general retinal anatomy including the topography of both the photoreceptor and ganglion cell populations and estimates the spatial resolving power (SRP) of the eye of the Mediterranean triplefin Tripterygion delaisi. Retinal wholemounts were prepared to systematically and quantitatively analyze photoreceptor and retinal ganglion cell (RGC) densities using design-based stereology. To further examine the retinal structure, we also used magnetic resonance imaging (MRI) and histological examination of retinal cross sections. Observations of the triplefin’s eyes revealed them to be highly mobile, allowing them to view the surroundings without body movements. A rostral aphakic gap and the elliptical shape of the eye extend its visual field rostrally and allow for a rostro-caudal accommodatory axis, enabling this species to focus on prey at close range. Single and twin cones dominate the retina and are consistently arranged in one of two regular patterns, which may enhance motion detection and color vision. The retina features a prominent, dorso-temporal, convexiclivate fovea with an average density of 104,400 double and 30,800 single cones per mm2, and 81,000 RGCs per mm2. Based on photoreceptor spacing, SRP was calculated to be between 6.7 and 9.0 cycles per degree. Location and resolving power of the fovea would benefit the detection and identification of small prey in the lower frontal region of the visual field.

The transformative power of the thumbnail image: Media logistics and infrastructural aesthetics

Thumbnail images are discreet, yet central navigational tools in increasingly complex visual information environments. Indeed, without thumbnail images there would be no image search: they are an inherent part of the information architecture of most digital information platforms. Yet, how might we understand the role of the thumbnail as an attention technology in the digital economy? And what kind of aesthetic does it produce? This paper examines the legal negotiations of the thumbnail image and the ensuing decision to conceptualize the thumbnail as a functional image against the cultural history of visual attention technologies and the aesthetics of their connective function. Such an endeavour, we propose, allows us to understand and appreciate the significant digital economy and particular aesthetic of the thumbnail image despite its apparent subtlety.