Search

Developmental myelination is a protracted process in the mammalian brain1. One theory for why oligodendrocytes mature so slowly posits that myelination may stabilize neuronal circuits and temper neuronal plasticity as animals age2-4. We tested this theory in the visual cortex, which has a well-defined critical period for experience-dependent neuronal plasticity5. During adolescence, visual experience modulated the rate of oligodendrocyte maturation in visual cortex. To determine whether oligodendrocyte maturation in turn regulates neuronal plasticity, we genetically blocked oligodendrocyte differentiation and myelination in adolescent mice. In adult mice lacking adolescent oligodendrogenesis, a brief period of monocular deprivation led to a significant decrease in visual cortex responses to the deprived eye, reminiscent of the plasticity normally restricted to adolescence. This enhanced functional plasticity was accompanied by a greater turnover of dendritic spines and coordinated reductions in spine size following deprivation. Furthermore, inhibitory synaptic transmission, which gates experience-dependent plasticity at the circuit level, was diminished in the absence of adolescent oligodendrogenesis. These results establish a critical role for oligodendrocytes in shaping the maturation and stabilization of cortical circuits and support the concept of developmental myelination acting as a functional brake on neuronal plasticity.

Three-dimensional perception in intelligent virtual and augmented reality (VR/AR) and autonomous vehicles (AV) applications is critical and attracting significant attention. The self-supervised monocular depth and ego-motion estimation serves as a more intelligent learning approach that provides the required scene depth and location for 3D perception. However, the existing self-supervised learning methods suffer from scale ambiguity, boundary blur, and imbalanced depth distribution, limiting the practical applications of VR/AR and AV. In this article, we propose a new self-supervised learning framework based on superpixel and normal constraints to address these problems. Specifically, we formulate a novel 3D edge structure consistency loss to alleviate the boundary blur of depth estimation. To address the scale ambiguity of estimated depth and ego-motion, we propose a novel surface normal network for efficient camera height estimation. The surface normal network is composed of a deep fusion module and a full-scale hierarchical feature aggregation module. Meanwhile, to realize the global smoothing and boundary discriminability of the predicted normal map, we introduce a novel fusion loss which is based on the consistency constraints of the normal in edge domains and superpixel regions. Experiments are conducted on several benchmarks, and the results illustrate that the proposed approach outperforms the state-of-the-art methods in depth, ego-motion, and surface normal estimation. [ABSTRACT FROM AUTHOR]

Data augmentation is widely recognized as an effective means of bolstering model robustness. However, when applied to monocular 3D object detection, non-geometric image augmentation neglects the critical link between the image and physical space, resulting in the semantic collapse of the extended scene. To address this issue, we propose two geometric-level data augmentation operators named Geometric-Copy-Paste (Geo-CP) and Geometric-Crop-Shrink (Geo-CS). Both operators introduce geometric consistency based on the principle of perspective projection, complementing the options available for data augmentation in monocular 3D. Specifically, Geo-CP replicates local patches by reordering object depths to mitigate perspective occlusion conflicts, and Geo-CS re-crops local patches for simultaneous scaling of distance and scale to unify appearance and annotation. These operations ameliorate the problem of class imbalance in the monocular paradigm by increasing the quantity and distribution of geometrically consistent samples. Experiments demonstrate that our geometric-level augmentation operators effectively improve robustness and performance in the KITTI and Waymo monocular 3D detection benchmarks. [ABSTRACT FROM AUTHOR]

A monocular ranging method for forward vehicles in intelligent driving is proposed. This method measures vehicle distance more accurately under the condition of a single camera and can estimate camera attitude in real-time. For the estimation of camera pitch and yaw angles, it is achieved using road vanishing points. The images collected by the camera are sequentially processed through the Roberts operator amplitude calculation, feature point extraction, feature line segment generation, road vanishing point voting, and estimation of camera attitude to obtain pitch and yaw angles. A distance estimation network was designed, which is divided into multiple levels based on image size and incorporates image feature, integrating vehicle grounding point, and vehicle width information, effectively improving ranging accuracy. Finally, validation was conducted on KITTI data, with a relative error (AbsRel) of 8.3%. Additionally, the TuSimple dataset and continuous driving scenarios were also validated, resulting in improved performance compared to previous algorithms. [ABSTRACT FROM AUTHOR]

Self-supervised monocular depth estimation has been widely applied in autonomous driving and automated guided vehicles. It offers the advantages of low cost and extended effective distance compared with alternative methods. However, like automated guided vehicles, devices with limited computing resources struggle to leverage state-of-the-art large model structures. In recent years, researchers have acknowledged this issue and endeavored to reduce model size. Model lightweight techniques aim to decrease the number of parameters while maintaining satisfactory performance. In this paper, to enhance the model's performance in lightweight scenarios, a novel approach to encompassing three key aspects is proposed: (1) utilizing LeakyReLU to involve more neurons in manifold representation; (2) employing large convolution for improved recognition of edges in lightweight models; (3) applying channel grouping and shuffling to maximize the model efficiency. Experimental results demonstrate that our proposed method achieves satisfactory outcomes on KITTI and Make3D benchmarks while having only 1.6M trainable parameters, representing a reduction of 27% compared with the previous smallest model, Lite-Mono-tiny, in monocular depth estimation. [ABSTRACT FROM AUTHOR]

Remotely Operated Vehicles (ROVs) are essential instruments in most industrial applications of subsea Inspection, Maintenance, and Repair (IMR). In IMR applications, especially in short-distance inspection and intervention operations, ROVs should be able to find its position and orientation with respect to underwater platforms or specific objects on the platform. Current work class ROVs are normally operated by two pilots. Improving the situational awareness and sensing capabilities is an essential necessity from industrial actors in such a challenging environment. Automated visual based interpretation shows promising results and essential tools toward higher level of autonomy within IMR and ROV operations. Motivated by the large amount of available information in ROV cameras, this article presents a pipeline for object detection, 6D pose estimation and tracking. The main contribution of the paper is its new detection procedure where the current tracker state is evaluated for acceptance or rejection of the new detection. The decomposed tasks of detection, 6D pose estimation and tracking in the proposed pipeline is tested on an open underwater dataset featuring pose annotated objects used for ROV intervention. The results show that the method is highly suitable for supervised autonomy and a step forward to autonomous operations in underwater IMR applications.

Deep learning models and algorithms facilitate relatively easier ways of hand pose estimation from monocular RGB images compared to traditional approaches. Despite this, a majority of available algorithms use multiple-stage models to perform hand pose estimation. Moreover, the single-stage methods are mainly limited to a single hand and it is difficult for them to scale to multiple hands. To this end, we propose an approach that takes the features of the saliency map extracted for hand region of interest (ROI) localization. An integrated network uses these features for pose estimation. This arrangement of layers forms an end-to-end pipeline that allows simultaneous pose estimation for multiple hands. The model is designed to run on multiple cores of CPU/GPU to independently perform inference for each detected hands’pose making possible faster inference and hence suitable for real-time applications. In addition, a new approach using grid-based design to estimate hand-keypoints position with high precision is also proposed. Both the proposed designs are validated on multiple datasets to prove their feasibility and effectiveness. The probability of the correct keypoint (PCK) value at threshold value of 0.2 is above 95% on the test sets from Interhand dataset and Rendered HandPose Dataset (RHD).

This survey provides a comprehensive overview of traditional techniques and deep learning-based methodologies for monocular visual odometry (VO), with a focus on displacement measurement applications. This paper outlines the fundamental concepts and general procedures for VO implementation, including feature detection, tracking, motion estimation, triangulation, and trajectory estimation. This paper also explores the research challenges inherent in VO implementation, including scale estimation and ground plane considerations. The scientific literature is rife with diverse methodologies aiming to overcome these challenges, particularly focusing on the problem of accurate scale estimation. This issue has been typically addressed through the reliance on knowledge regarding the height of the camera from the ground plane and the evaluation of feature movements on that plane. Alternatively, some approaches have utilized additional tools, such as LiDAR or depth sensors. This survey of approaches concludes with a discussion of future research challenges and opportunities in the field of monocular visual odometry. [ABSTRACT FROM AUTHOR]

Purpose: We aimed to identify ocular comorbidities and reasons of blindness in monocular patients and to compare visual outcomes of cataract surgery between monocular and binocular patients. Methods: A single-center case–control study was conducted between November 2011 and May 2019 to compare consecutive series of patients needing cataract surgery in Strasbourg University Hospitals, France. Cases were patients with permanent monocular vision loss. Controls were binocularly sighted patients. All patients underwent cataract surgery using phacoemulsification technique. Chart analysis included demographic data, medical history, and surgical determinants data. Student's t tests and Fisher’s exact tests were the main methods used for statistical analysis. Results: Each group included 80 patients. The mean age at the time of surgery was significantly higher in monocular than binocular patients (77 vs. 71 years, p < 0.001). Thirty-two monocular patients (40%) had ocular comorbidities, compared to only 19 (23%) in the control group (p < 0.05). The leading cause of monocular status was amblyopia caused by strabismus (22 patients, 27.5%). Age-related macular degeneration, open-angle glaucoma, and diabetic retinopathy were the three main ocular comorbidities that were observed in the monocular group. Monocular patients had significantly lower visual acuity than the control group (p < 0.01) before and after cataract surgery. Conversely, improvement in visual acuity after surgery was not statistically different between groups (p = 0.054). There was no statistically significant difference in the rate of surgical complications between groups (p = 0.622). Conclusions: This study illustrates that cataract surgery in monocular patients is not more complicated than in binocular patients, but that it is significantly delayed. [ABSTRACT FROM AUTHOR]

Existing methods for 3D tracking from monocular RGB videos predominantly consider articulated and rigid objects (e.g., two hands or humans interacting with rigid environments). Modelling dense non-rigid object deformations in this setting (e.g. when hands are interacting with a face), remained largely unaddressed so far, although such effects can improve the realism of the downstream applications such as AR/VR, 3D virtual avatar communications, and character animations. This is due to the severe ill-posedness of the monocular view setting and the associated challenges (e.g., in acquiring a dataset for training and evaluation or obtaining the reasonable non-uniform stiffness of the deformable object). While it is possible to naïvely track multiple non-rigid objects independently using 3D templates or parametric 3D models, such an approach would suffer from multiple artefacts in the resulting 3D estimates such as depth ambiguity, unnatural intra-object collisions and missing or implausible deformations. Hence, this paper introduces the first method that addresses the fundamental challenges depicted above and that allows tracking human hands interacting with human faces in 3D from single monocular RGB videos. We model hands as articulated objects inducing non-rigid face deformations during an active interaction. Our method relies on a new hand-face motion and interaction capture dataset with realistic face deformations acquired with a markerless multi-view camera system. As a pivotal step in its creation, we process the reconstructed raw 3D shapes with position-based dynamics and an approach for non-uniform stiffness estimation of the head tissues, which results in plausible annotations of the surface deformations, hand-face contact regions and head-hand positions. At the core of our neural approach are a variational auto-encoder supplying the hand-face depth prior and modules that guide the 3D tracking by estimating the contacts and the deformations. Our final 3D hand and face reconstructions are realistic and more plausible compared to several baselines applicable in our setting, both quantitatively and qualitatively. https://vcai.mpi-inf.mpg.de/projects/Decaf [ABSTRACT FROM AUTHOR]

A monocular ranging method for forward vehicles in intelligent driving is proposed. This method measures vehicle distance more accurately under the condition of a single camera and can estimate camera attitude in real-time. For the estimation of camera pitch and yaw angles, it is achieved using road vanishing points. The images collected by the camera are sequentially processed through the Roberts operator amplitude calculation, feature point extraction, feature line segment generation, road vanishing point voting, and estimation of camera attitude to obtain pitch and yaw angles. A distance estimation network was designed, which is divided into multiple levels based on image size and incorporates image feature, integrating vehicle grounding point, and vehicle width information, effectively improving ranging accuracy. Finally, validation was conducted on KITTI data, with a relative error (AbsRel) of 8.3%. Additionally, the TuSimple dataset and continuous driving scenarios were also validated, resulting in improved performance compared to previous algorithms.

We present a synthetic augmentation approach towards improving monocular face presentation–attack–detection (PAD) robustness to real-world noise additions. Face PAD algorithms secure authentication systems against spoofing attacks, such as pictures, videos, and 2D-inspired masks. Best-in-class PAD methods typically use 3D imagery, but these can be expensive. To reduce application cost, there is a growing field investigating monocular algorithms that detect facial artifacts. These approaches work well in laboratory conditions, but can be sensitive to the imaging environment (e.g., sensor noise, dynamic lighting, etc.). The ideal solution for noise robustness is training under all expected conditions; however, this is time consuming and expensive. Instead, we propose that physics-informed noise-augmentations can pragmatically achieve robustness. Our toolbox contains twelve sensor and lighting effect generators. We demonstrate that our toolbox generates more robust PAD features than popular augmentation methods in noisy test-evaluations. We also observe that the toolbox improves accuracy on clean test data, suggesting that it inherently helps discern spoof artifacts from imaging artifacts. We validate this hypothesis through an ablation study, where we remove liveliness pairs (e.g., live or spoof imagery only for participants) to identify how much real data can be replaced with synthetic augmentations. We demonstrate that using these noise augmentations allows us to achieve better test accuracy while only requiring 30% of participants to be fully imaged under all conditions. These findings indicate that synthetic noise augmentations are a great way to improve PAD, addressing noise robustness while simplifying data collection. [ABSTRACT FROM AUTHOR]

Copyright of Laser Technology is the property of Gai Kan Bian Wei Hui and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Purpose: Abnormalities in the bladder wall require careful investigation regarding type, spatial position and invasiveness. Construction of a 3-D model of the bladder is helpful to ensure adequate coverage of the scanning procedure, quantitative comparison of bladder wall textures between successive sessions and finding back previously discovered abnormalities. Methods: Videos of both an in vivo bladder and a textured bladder phantom were acquired. Structure-from-motion and bundle adjustment algorithms were used to construct a 3-D point cloud, approximate it by a surface mesh, texture it with the back-projected camera frames and draw the corresponding 2-D atlas. Reconstructions of successive sessions were compared; those of the bladder phantom were co-registered, transformed using 3-D thin plate splines and post-processed to highlight significant changes in texture. Results: The reconstruction algorithms of the presented workflow were able to construct 3-D models and corresponding 2-D atlas of both the in vivo bladder and the bladder phantom. For the in vivo bladder the portion of the reconstructed surface area was 58% and 79% for the pre- and post-operative scan, respectively. For the bladder phantom the full surface was reconstructed and the mean reprojection error was 0.081 mm (range 0–0.79 mm). In inter-session comparison the changes in texture were correctly indicated for all six locations. Conclusion: The proposed proof of concept was able to perform 3-D and 2-D reconstruction of an in vivo bladder wall based on a set of monocular images. In a phantom study the computer vision algorithms were also effective in co-registering reconstructions of successive sessions and highlighting texture changes between sessions. These techniques may be useful for detecting, monitoring and revisiting suspicious lesions. [ABSTRACT FROM AUTHOR]

The mammalian visual cortex contains multiple retinotopically defined areas that process distinct features of the visual scene. Little is known about what guides the functional differentiation of visual cortical areas during development. Recent studies in mice have revealed that visual input from the two eyes provides spatiotemporally distinct signals to primary visual cortex (V1), such that contralateral eye-dominated V1 neurons respond to higher spatial frequencies than ipsilateral eye-dominated neurons. To test whether binocular visual input drives the differentiation of visual cortical areas, we used two-photon calcium imaging to characterize the effects of juvenile monocular deprivation (MD) on the responses of neurons in V1 and two higher visual areas, LM (lateromedial) and PM (posteromedial). In adult mice of either sex, we find that MD prevents the emergence of distinct spatiotemporal tuning in V1, LM, and PM. We also find that, within each of these areas, MD reorganizes the distinct spatiotemporal tuning properties driven by the two eyes. Moreover, we find a relationship between speed tuning and ocular dominance in all three areas that MD preferentially disrupts in V1, but not in LM or PM. Together, these results reveal that balanced binocular vision during development is essential for driving the functional differentiation of visual cortical areas. The higher visual areas of mouse visual cortex may provide a useful platform for investigating the experience-dependent mechanisms that set up the specialized processing within neocortical areas during postnatal development.SIGNIFICANCE STATEMENT Little is known about the factors guiding the emergence of functionally distinct areas in the brain. Using in vivo Ca2+ imaging, we recorded visually evoked activity from cells in V1 and higher visual areas LM (lateromedial) and PM (posteromedial) of mice. Neurons in these areas normally display distinct spatiotemporal tuning properties. We found that depriving one eye of normal input during development prevents the functional differentiation of visual areas. Deprivation did not disrupt the degree of speed tuning, a property thought to emerge in higher visual areas. Thus, some properties of visual cortical neurons are shaped by binocular experience, while others are resistant. Our study uncovers the fundamental role of binocular experience in the formation of distinct areas in visual cortex.

Myopia is induced when a growing eye wears a diffuser that deprives it of detailed spatial vision (form deprivation, FD). In chickens with optic nerve section (ONS), FD myopia still occurs, suggesting that the signals underlying myopia reside within the eye. As avian eyes differ from mammals, we asked whether local mechanisms also underlie FD myopia in a mammalian model. Young guinea pigs underwent either sham surgery followed by FD (SHAM + FD, n = 7); or ONS followed by FD (ONS + FD, n = 7); or ONS without FD (ONS, n = 9). FD was initiated 3 days after surgery with a diffuser that was worn on the surgically treated eye for 14 days. Animals with ONS + FD developed -8.9 D of relative myopia and elongated by 135 μm more than in their untreated eyes after 2 weeks of FD. These changes were significantly greater than those in SHAM + FD animals (-5.5 D and 40 μm of elongation after 14 days of FD), and reflected exaggerated elongation of the posterior vitreous chamber. The myopia reversed when FD was discontinued, despite ONS, but eyes did not recover back to normal (30 days after surgery, ONS + FD eyes still retained -3 D of relative myopia when SHAM+FD animals had returned to normal). No long-term residual myopia was present after ONS alone, ruling out a surgical artifact. Although the gross mechanism signaling myopic ocular growth and its recovery in the young mammalian eye does not require an intact optic nerve, its fine-tuning is disrupted by ONS.

Purpose: To evaluate the determinants affecting the quality of life in monocular glaucoma patients using the Indian vision function questionnaire. Methods: In this prospective cross-sectional study, total of 196 patients were divided into two groups: cases and controls. Indian Vision Function Questionnaire (IND-VFQ) was administered and analyzed. One hundred twenty-nine (58.6%) patients who had lost their vision in one eye due to glaucoma were included as cases and 67 (30.4%) patients who had lost their vision due to other causes were taken as controls. Results: Median composite score of subscales was 54.62 (29.7-74.7) in group 1 and 45.38 (23.7-76.7) in group 2. The psychosocial impact scale was the most affected scale, the median scores were 33.02 (0 to 60.0) and 19.07 (0 to 53.0) in groups 1 and 2, respectively. Among all dimensions of IND-VFQ, the highest score was for color vision 100.0 (0-100.0) and 100.0 (0-100.0), and the lowest median score was found in mental health and dependency in both the groups. Multiple linear regression analysis demonstrated that visual acuity was associated with a low score (P < 0.001). Female gender was significantly associated with the overall score in the univariate model (P = 0.006). Conclusion: Monocular glaucoma patients have a poor general and vision-related quality of life. Depression associated with monocularity and the perception of dependency and being a burden on their family members greatly impacted the mental health of the participants. [ABSTRACT FROM AUTHOR]

High acuity stereopsis emerges during an early postnatal critical period when binocular neurons in the primary visual cortex sharpen their receptive field tuning properties. We find that this sharpening is achieved by dismantling the binocular circuit present at critical period onset and building it anew. Longitudinal imaging of receptive field tuning (e.g., orientation selectivity) of thousands of neurons reveals that most binocular neurons present in layer 2/3 at critical period onset are poorly tuned and are rendered monocular. In parallel, new binocular neurons are established by conversion of well-tuned monocular neurons as they gain matched input from the other eye. These improvements in binocular tuning in layer 2/3 are not inherited from layer 4 but are driven by the experience-dependent sharpening of ipsilateral eye responses. Thus, vision builds a new and more sharply tuned binocular circuit in layer 2/3 by cellular exchange and not by refining the original circuit.

Study of the neural deficits caused by mismatched binocular vision in early childhood has predominantly focused on circuits in the primary visual cortex (V1). Recent evidence has revealed that neurons in mouse dorsolateral geniculate nucleus (dLGN) can undergo rapid ocular dominance plasticity following monocular deprivation (MD). It remains unclear, however, whether the long-lasting deficits attributed to MD during the critical period originate in the thalamus. Using in vivo two-photon Ca2+ imaging of dLGN afferents in superficial layers of V1 in female and male mice, we demonstrate that 14 d MD during the critical period leads to a chronic loss of binocular dLGN inputs while sparing response strength and spatial acuity. Importantly, MD leads to profoundly mismatched visual tuning properties in remaining binocular dLGN afferents. Furthermore, MD impairs binocular modulation, reducing facilitation of responses of both binocular and monocular dLGN inputs during binocular viewing. As predicted by our findings in thalamic inputs, Ca2+ imaging from V1 neurons revealed spared spatial acuity but impaired binocularity in L4 neurons. V1 L2/3 neurons in contrast displayed deficits in both binocularity and spatial acuity. Our data demonstrate that critical-period MD produces long-lasting disruptions in binocular integration beginning in early binocular circuits in dLGN, whereas spatial acuity deficits first arise from circuits further downstream in V1. Our findings indicate that the development of normal binocular vision and spatial acuity depend upon experience-dependent refinement of distinct stages in the mammalian visual system.SIGNIFICANCE STATEMENT Abnormal binocular vision and reduced acuity are hallmarks of amblyopia, a disorder that affects 2%-5% of the population. It is widely thought that the neural deficits underlying amblyopia begin in the circuits of primary visual cortex. Using in vivo two-photon calcium imaging of thalamocortical axons in mice, we show that depriving one eye of input during a critical period in development chronically impairs binocular integration in thalamic inputs to primary visual cortex. In contrast, visual acuity is spared in thalamic inputs. These findings shed new light on the role for developmental mechanisms in the thalamus in establishing binocular vision and may have critical implications for amblyopia.

Purpose: To evaluate the determinants affecting the quality of life in monocular glaucoma patients using the Indian vision function questionnaire. Methods: In this prospective cross-sectional study, total of 196 patients were divided into two groups: cases and controls. Indian Vision Function Questionnaire (IND-VFQ) was administered and analyzed. One hundred twenty-nine (58.6%) patients who had lost their vision in one eye due to glaucoma were included as cases and 67 (30.4%) patients who had lost their vision due to other causes were taken as controls. Results: Median composite score of subscales was 54.62 (29.7–74.7) in group 1 and 45.38 (23.7–76.7) in group 2. The psychosocial impact scale was the most affected scale, the median scores were 33.02 (0 to 60.0) and 19.07 (0 to 53.0) in groups 1 and 2, respectively. Among all dimensions of IND-VFQ, the highest score was for color vision 100.0 (0–100.0) and 100.0 (0–100.0), and the lowest median score was found in mental health and dependency in both the groups. Multiple linear regression analysis demonstrated that visual acuity was associated with a low score (P < 0.001). Female gender was significantly associated with the overall score in the univariate model (P = 0.006). Conclusion: Monocular glaucoma patients have a poor general and vision-related quality of life. Depression associated with monocularity and the perception of dependency and being a burden on their family members greatly impacted the mental health of the participants.

AIM:To evaluate the underlying aetiology and clinical characteristics of retinal detachment(RD)in school-age pediatric monocular RD.METHODS:Patients with RD and contralateral blind(monocular RD)aged 7-14 years, from November 2015 to May 2021 in our hospital were retrospectively reviewed. Demographic characteristics and etiology of RD, clinical type, surgical modality, type of intraocular tamponade, pre-and postoperative visual and anatomical outcomes were recorded and evaluated.RESULTS: There were 27 children(27 eyes)with monocular RD at least 6mo follow-up. The average age at presentation was 10.63±2.30 years. Familial exudative vitreoretinopathy(FEVR)(11/27, 41%), postoperative congenital glaucoma(6/27, 22%)and Stickler syndrome(3/27, 11%)were main underlying etiologies. Among them, rhegmatogenous retinal detachment(RRD)comprised 78%(21/27)of the patients, of which 81% patients(17/21)had proliferative vitreoretinopathy(PVR)C3 or worse. Pars plana vitrectomy(PPV)was done in 85%(23/27)of the patients, of which 83%(19/23)received silicone oil tamponade. Best corrected visual acuity(BCVA, LogMAR)worse than 1.7 was seen in 78%(21/27)of the patients at final visit, and 82%(22/27)had reattached retina, but 41%(11/27)of the patients remained status of silicone oil tamponade at last visit.CONCLUSION:School-age pediatric monocular RD is often associated with underlying congenital or hereditary conditions, and often presented with severe RD and severe PVR reaction which needed vitrectomy combined with silicon oil tamponade, and with poor visual and anatomical short-term prognosis.

Object pose detection and tracking has recently attracted increasing attention due to its wide applications in many areas, such as autonomous driving, robotics, and augmented reality. Among methods for object pose detection and tracking, deep learning is the most promising one that has shown better performance than others. However, survey study about the latest development of deep learning-based methods is lacking. Therefore, this study presents a comprehensive review of recent progress in object pose detection and tracking that belongs to the deep learning technical route. To achieve a more thorough introduction, the scope of this study is limited to methods taking monocular RGB/RGBD data as input and covering three kinds of major tasks: instance-level monocular object pose detection, category-level monocular object pose detection, and monocular object pose tracking. In our work, metrics, datasets, and methods of both detection and tracking are presented in detail. Comparative results of current state-of-the-art methods on several publicly available datasets are also presented, together with insightful observations and inspiring future research directions. [ABSTRACT FROM AUTHOR]

We introduce Vision UFormer (ViUT), a novel deep neural long-range monocular depth estimator. The input is an RGB image, and the output is an image that stores the absolute distance of the object in the scene as its per-pixel values. ViUT consists of a Transformer encoder and a ResNet decoder combined with the UNet style of skip connections. It is trained on 1M images across ten datasets in a staged regime that starts with easier-to-predict data such as indoor photographs and continues to more complex long-range outdoor scenes. We show that ViUT provides comparable results for normalized relative distances and short-range classical datasets such as NYUv2 and KITTI. We further show that it successfully estimates absolute long-range depth in meters. We validate ViUT on a wide variety of long-range scenes showing its high estimation capabilities with a relative improvement of up to 23%. Absolute depth estimation finds application in many areas, and we show its usability in image composition, range annotation, defocus, and scene reconstruction. Our models are available at cphoto.fit.vutbr.cz/viut. [Display omitted] • Vision UFormer: Dense depth prediction model combining Vision Transformer with a UNet. • Staged Training: Moving from easier to difficult data allows successful training. • Predictor reaches SOTA results, surpassing others in long-range natural environments • Depths are used for further applications, e.g., scene reconstruction or manipulation. [ABSTRACT FROM AUTHOR]

For an advanced driver assistance system (ADAS) that automatically controls and operates the vehicle system in order to realize a safe and secure society in which vehicles and people coexist, we examined a method to realize stereo vision with an inexpensive configuration in a monocular camera. Generally, a monocular camera can construct an ADAS system at low cost, but there is a problem in terms of distance measurement accuracy. On the other hand, a stereo camera that can measure the distance using two monocular cameras with the same mechanism as the human eye has high distance measurement performance, which is advantageous for the sophistication of the ADAS system, but the cost increases. It becomes an issue. Therefore, in this research, we investigated an inexpensive stereo camera and devised a method to realize stereoscopic vision by dividing the field of view of a monocular camera using two reflective mirrors with different postures. We built a prototype and evaluated this method, demonstrating that it can measure a distance of 30 m with an error of 12%. [ABSTRACT FROM AUTHOR]

This survey provides a comprehensive overview of traditional techniques and deep learning-based methodologies for monocular visual odometry (VO), with a focus on displacement measurement applications. This paper outlines the fundamental concepts and general procedures for VO implementation, including feature detection, tracking, motion estimation, triangulation, and trajectory estimation. This paper also explores the research challenges inherent in VO implementation, including scale estimation and ground plane considerations. The scientific literature is rife with diverse methodologies aiming to overcome these challenges, particularly focusing on the problem of accurate scale estimation. This issue has been typically addressed through the reliance on knowledge regarding the height of the camera from the ground plane and the evaluation of feature movements on that plane. Alternatively, some approaches have utilized additional tools, such as LiDAR or depth sensors. This survey of approaches concludes with a discussion of future research challenges and opportunities in the field of monocular visual odometry.

We sought to determine whether a putative lower-visual field (loVF) advantage for projections to the visuomotor networks of the dorsal visual pathway influences online reaching control. Participants reached to 3-dimensional depth targets presented in the loVF and upper-visual field (upVF) in binocular and monocular visual conditions, and when online vision was available (i.e., closed-loop) or unavailable (i.e., open-loop). To examine the degree to which responses were controlled online we computed the proportion of variance (R2) explained by the spatial position of the limb at distinct stages in the reaching trajectory relative to a response's ultimate movement endpoint. Results showed that binocular and closed-loop reaches exhibited shorter movement times and more online corrections (i.e., smaller R2 values) than their monocular and open-loop counterparts. Notably, however, loVF and upper-visual field reaches exhibited equivalent performance metrics across all experimental conditions. Accordingly, results provide no evidence of a loVF advantage for online reaching control to 3-dimensional targets. [ABSTRACT FROM AUTHOR]

The developing brain can respond quickly to altered sensory experience by circuit reorganization. During a critical period in early life, neurons in the primary visual cortex rapidly lose responsiveness to an occluded eye and come to respond better to the open eye. While physiological and some of the molecular mechanisms of this process have been characterized, its structural basis, except for the well-known changes in the thalamocortical projection, remains obscure. To elucidate the relationship between synaptic remodeling and functional changes during this experience-dependent process, we used 2-photon microscopy to image synaptic structures of sparsely labeled layer 2/3 neurons in the binocular zone of mouse primary visual cortex. Anatomical changes at presynaptic and postsynaptic sites in mice undergoing monocular visual deprivation (MD) were compared to those in control mice with normal visual experience. We found that postsynaptic spines remodeled quickly in response to MD, with neurons more strongly dominated by the deprived eye losing more spines. These postsynaptic changes parallel changes in visual responses during MD and their recovery after restoration of binocular vision. In control animals with normal visual experience, the formation of presynaptic boutons increased during the critical period and then declined. MD affected bouton formation, but with a delay, blocking it after 3 d. These findings reveal intracortical anatomical changes in cellular layers of the cortex that can account for rapid activity-dependent plasticity.

Featured Application: Face-recognition for monocular systems (e.g., phones and buildings). This paper presents a novel material spectroscopy approach to facial presentation–attack–defense (PAD). Best-in-class PAD methods typically detect artifacts in the 3D space. This paper proposes similar features can be achieved in a monocular, single-frame approach by using controlled light. A mathematical model is produced to show how live faces and their spoof counterparts have unique reflectance patterns due to geometry and albedo. A rigorous dataset is collected to evaluate this proposal: 30 diverse adults and their spoofs (paper-mask, display-replay, spandex-mask and COVID mask) under varied pose, position, and lighting for 80,000 unique frames. A panel of 13 texture classifiers are then benchmarked to verify the hypothesis. The experimental results are excellent. The material spectroscopy process enables a conventional MobileNetV3 network to achieve 0.8% average-classification-error rate, outperforming the selected state-of-the-art algorithms. This demonstrates the proposed imaging methodology generates extremely robust features. [ABSTRACT FROM AUTHOR]

• Birth to surgery interval of 4–6 months crucial for stereopsis in congenital cataract. • Better stereopsis in bilateral congenital cataract than unilateral counterpart. • Stereopsis drops by 9%–16% in congenital cataracts with pre-existing strabismus. • No association found between strabismus and stereopsis in developmental cataracts. We estimated the proportion of children with stereopsis following surgery in congenital and developmental cataracts by systematic review and meta-analysis and also considered the factors influencing stereopsis, such as intervention age and presence of strabismus. Stereopsis is directly related to quality of life, and investigating its levels following cataract surgery in children may help decide the right time to intervene, particularly in the context of brain plasticity. We conducted a systematic literature search using Scopus, PubMed, and Web of Science and found 25 case series, 3 cohorts, and 3 clinical trial studies from 1/1/1995 to 31/12/2020. Study-specific proportions of stereopsis from 923 children were pooled using a random-effects model, and stratified analyses were conducted based on intervention age and pre-existing strabismus as a confounder. We appraised the risk of bias using tools published by National Institutes of Health and evaluated publication bias with funnel plots and the Egger test. The pooled proportions of stereopsis based on 8 unilateral and 6 bilateral congenital cataract studies were 0.37 (95% CIs: [0.24, 0.53]) and 0.45 (95% CIs: [0.24,0.68]) when patients with preexisting strabismus were excluded as a confounder. When the intervention age was ≤6 months, proportions in unilateral congenital cataract group significantly increased to 0.52 (95% CIs: [0.37, 0.66]; P = 0.49) compared to 0.26 (95% CIs: [0.14, 0.44]; P = 0.16) otherwise. A similar increase in proportions was found when intervention age ≤4 months. In both unilateral and bilateral congenital cataract groups, proportions increased significantly when the confounder was excluded. Overall, proportions in bilateral congenital cataracts were significantly greater than unilateral cases (irrespective of confounder). Eight unilateral and 5 bilateral developmental cataract studies resulted in pooled proportions of 0.62 (95% CIs: [0.27, 0.88] and 0.82 (95% CIs: [0.4, 0.97]), respectively. Although proportions for bilateral developmental cataracts were greater than unilateral cataracts (irrespective of confounder), results were not statistically significant. Finally, proportions in unilateral developmental cataracts were significantly greater than unilateral congenital cataracts (Z = 7.413, P = 6.173694e−14). We conclude that surgical intervention within first 4–6 months can significantly affect postoperative outcomes in unilateral congenital cataracts. Analysis of existing data does not show a significant effect of intervention age on stereopsis outcomes for developmental cataracts. [ABSTRACT FROM AUTHOR]

We present a synthetic augmentation approach towards improving monocular face presentation–attack–detection (PAD) robustness to real-world noise additions. Face PAD algorithms secure authentication systems against spoofing attacks, such as pictures, videos, and 2D-inspired masks. Best-in-class PAD methods typically use 3D imagery, but these can be expensive. To reduce application cost, there is a growing field investigating monocular algorithms that detect facial artifacts. These approaches work well in laboratory conditions, but can be sensitive to the imaging environment (e.g., sensor noise, dynamic lighting, etc.). The ideal solution for noise robustness is training under all expected conditions; however, this is time consuming and expensive. Instead, we propose that physics-informed noise-augmentations can pragmatically achieve robustness. Our toolbox contains twelve sensor and lighting effect generators. We demonstrate that our toolbox generates more robust PAD features than popular augmentation methods in noisy test-evaluations. We also observe that the toolbox improves accuracy on clean test data, suggesting that it inherently helps discern spoof artifacts from imaging artifacts. We validate this hypothesis through an ablation study, where we remove liveliness pairs (e.g., live or spoof imagery only for participants) to identify how much real data can be replaced with synthetic augmentations. We demonstrate that using these noise augmentations allows us to achieve better test accuracy while only requiring 30% of participants to be fully imaged under all conditions. These findings indicate that synthetic noise augmentations are a great way to improve PAD, addressing noise robustness while simplifying data collection.

The monocular simultaneous localization and mapping (SLAM) system is one of the most important among all visual or visual-inertial SLAM (VSLAM or VISLAM) systems due to its low cost, easy calibration and identification. Initialization is always crucial to bootstrap the monocular SLAM system. With the rapid growth of some quick-start required SLAM applications, e.g., augmented reality (AR) and unmanned aerial vehicles (UAVs), devising faster and more accurate initialization has become a central problem. Traditional initialization uses the first two frames to create landmarks and then computes camera poses for the subsequential frames, using a 3D-to-2D perspective-n-points (PnP) mechanism. In this paper, we propose a novel three-frame-only (3FO) initialization approach for the monocular SLAM system, which consists of two steps. In the first step, we use the first two frames to preinitialize poses and landmarks, and in the second step, we use the second and third frames to improve the preinitialization by using the scale consistency of the landmarks generated in the first step to filter out outliers and using inliers to generate more robust landmarks. In both steps, we use the pretrained multilayer perceptron (MLP) combined with homotopy continuation to solve the essential matrices. Finally, we perform a global bundle adjustment (BA) to refine the three camera poses and all the created landmarks. The proposed 3FO initialization approach is evaluated experimentally on the EuRoC benchmark data set with the initialization time and trajectory metrics. The results show that, compared to the traditional ORB-SLAM2 initialization, the 3FO approach reduces the initialization time by 5 times and improves the accuracy by 36.7% on average.

Peters and Lau (eLife, 4, e09651, 2015) found that when criterion bias is controlled for, there is no evidence for unconscious visual perception in normal observers, in the sense that they cannot directly discriminate a target above chance without knowing it. One criticism of that study is that the visual suppression method used, forward and backward masking (FBM), may be too blunt in the way it interferes with visual processing to allow for unconscious forced-choice discrimination. To investigate this question, we compared FBM directly to continuous flash suppression (CFS) in a two-interval forced-choice task. Although CFS is popular, and may be thought of as a more powerful visual suppression technique, we found no difference in the degree of perceptual impairment between the two suppression types. To the extent that CFS impairs perception, both objective discrimination and subjective awareness are impaired to similar degrees under FBM. This pattern was consistently observed across three experiments in which various experimental parameters were varied. These findings provide evidence for an ongoing debate about unconscious perception: normal observers cannot perform forced-choice discrimination tasks unconsciously.

When occlusion and binocular disparity cues conflict, what visual features determine how they combine? Sensory cues, such as T-junctions, have been suggested to be necessary for occlusion to influence stereoscopic depth perception. Here we show that illusory occlusion, with no retinal sensory cues, interacts with binocular disparity when perceiving depth. We generated illusory occlusion using stimuli filled in across the retinal blind spot. Observers viewed two bars forming a cross with the intersection positioned within the blind spot. One of the bars was presented binocularly with a disparity signal; the other was presented monocularly, extending through the blind spot, with no defined disparity. When the monocular bar was perceived as filled in through the blind spot, it was perceived as occluding the binocular bar, generating illusory occlusion. We found that this illusory occlusion influenced perceived stereoscopic depth: depth estimates were biased to be closer or farther, depending on whether a bar was perceived as in front of or behind the other bar, respectively. Therefore, the perceived relative depth position, based on filling-in cues, set boundaries for interpreting metric stereoscopic depth cues. This suggests that filling-in can produce opaque surface representations that can trump other depth cues such as disparity.

In order to improve the initialization robustness of visual inertial SLAM, the complementarity of the optical flow method and the feature-based method can be used in vision data processing. The parallel initialization method is proposed, where the optical flow inertial initialization and the monocular feature-based initialization are carried out at the same time. After the initializations, the state estimation results are jointly optimized by bundle adjustment. The proposed method retains more mapping information, and correspondingly is more adaptable to the initialization scene. It is found that the initialization map constructed by the proposed method features a comparable accuracy to the one constructed by ORB-SLAM3 in monocular inertial mode. Since the online extrinsic parameter estimation can be realized by the proposed method, it is considered better than ORB-SLAM3 in the aspect of portability. By the experiments performed on the benchmark dataset EuRoC, the effectiveness and robustness of the proposed method are validated. [ABSTRACT FROM AUTHOR]

Detecting and localizing objects in the real 3D space, which plays a crucial role in scene understanding, is particularly challenging given only a monocular image due to the geometric information loss during imagery projection. We propose MonoGRNet for the amodal 3D object detection from a monocular image via geometric reasoning in both the observed 2D projection and the unobserved depth dimension. MonoGRNet decomposes the monocular 3D object detection task into four sub-tasks including 2D object detection, instance-level depth estimation, projected 3D center estimation and local corner regression. The task decomposition significantly facilitates the monocular 3D object detection, allowing the target 3D bounding boxes to be efficiently predicted in a single forward pass, without using object proposals, post-processing or the computationally expensive pixel-level depth estimation utilized by previous methods. In addition, MonoGRNet flexibly adapts to both fully and weakly supervised learning, which improves the feasibility of our framework in diverse settings. Experiments are conducted on KITTI, Cityscapes and MS COCO datasets. Results demonstrate the promising performance of our framework in various scenarios. [ABSTRACT FROM AUTHOR]