Pothole Detection Based on Disparity Transformation and Road Surface Modeling.

Pothole detection is one of the most important tasks for road maintenance. Computer vision approaches are generally based on either 2D road image analysis or 3D road surface modeling. However, these two categories are always used independently. Furthermore, the pothole detection accuracy is still far from satisfactory. Therefore, in this paper, we present a robust pothole detection algorithm that is both accurate and computationally efficient. A dense disparity map is first transformed to better distinguish between damaged and undamaged road areas. To achieve greater disparity transformation efficiency, golden section search and dynamic programming are utilized to estimate the transformation parameters. Otsu's thresholding method is then used to extract potential undamaged road areas from the transformed disparity map. The disparities in the extracted areas are modeled by a quadratic surface using least squares fitting. To improve disparity map modeling robustness, the surface normal is also integrated into the surface modeling process. Furthermore, random sample consensus is utilized to reduce the effects caused by outliers. By comparing the difference between the actual and modeled disparity maps, the potholes can be detected accurately. Finally, the point clouds of the detected potholes are extracted from the reconstructed 3D road surface. The experimental results show that the successful detection accuracy of the proposed system is around 98.7% and the overall pixel-level accuracy is approximately 99.6%.

On the impact of Citizen Science-derived data quality on deep learning based classification in marine images.

The evaluation of large amounts of digital image data is of growing importance for biology, including for the exploration and monitoring of marine habitats. However, only a tiny percentage of the image data collected is evaluated by marine biologists who manually interpret and annotate the image contents, which can be slow and laborious. In order to overcome the bottleneck in image annotation, two strategies are increasingly proposed: "citizen science" and "machine learning". In this study, we investigated how the combination of citizen science, to detect objects, and machine learning, to classify megafauna, could be used to automate annotation of underwater images. For this purpose, multiple large data sets of citizen science annotations with different degrees of common errors and inaccuracies observed in citizen science data were simulated by modifying "gold standard" annotations done by an experienced marine biologist. The parameters of the simulation were determined on the basis of two citizen science experiments. It allowed us to analyze the relationship between the outcome of a citizen science study and the quality of the classifications of a deep learning megafauna classifier. The results show great potential for combining citizen science with machine learning, provided that the participants are informed precisely about the annotation protocol. Inaccuracies in the position of the annotation had the most substantial influence on the classification accuracy, whereas the size of the marking and false positive detections had a smaller influence.

A framework for the development of a global standardised marine taxon reference image database (SMarTaR-ID) to support image-based analyses.

Video and image data are regularly used in the field of benthic ecology to document biodiversity. However, their use is subject to a number of challenges, principally the identification of taxa within the images without associated physical specimens. The challenge of applying traditional taxonomic keys to the identification of fauna from images has led to the development of personal, group, or institution level reference image catalogues of operational taxonomic units (OTUs) or morphospecies. Lack of standardisation among these reference catalogues has led to problems with observer bias and the inability to combine datasets across studies. In addition, lack of a common reference standard is stifling efforts in the application of artificial intelligence to taxon identification. Using the North Atlantic deep sea as a case study, we propose a database structure to facilitate standardisation of morphospecies image catalogues between research groups and support future use in multiple front-end applications. We also propose a framework for coordination of international efforts to develop reference guides for the identification of marine species from images. The proposed structure maps to the Darwin Core standard to allow integration with existing databases. We suggest a management framework where high-level taxonomic groups are curated by a regional team, consisting of both end users and taxonomic experts. We identify a mechanism by which overall quality of data within a common reference guide could be raised over the next decade. Finally, we discuss the role of a common reference standard in advancing marine ecology and supporting sustainable use of this ecosystem.

MAIA-A machine learning assisted image annotation method for environmental monitoring and exploration.

Digital imaging has become one of the most important techniques in environmental monitoring and exploration. In the case of the marine environment, mobile platforms such as autonomous underwater vehicles (AUVs) are now equipped with high-resolution cameras to capture huge collections of images from the seabed. However, the timely evaluation of all these images presents a bottleneck problem as tens of thousands or more images can be collected during a single dive. This makes computational support for marine image analysis essential. Computer-aided analysis of environmental images (and marine images in particular) with machine learning algorithms is promising, but challenging and different to other imaging domains because training data and class labels cannot be collected as efficiently and comprehensively as in other areas. In this paper, we present Machine learning Assisted Image Annotation (MAIA), a new image annotation method for environmental monitoring and exploration that overcomes the obstacle of missing training data. The method uses a combination of autoencoder networks and Mask Region-based Convolutional Neural Network (Mask R-CNN), which allows human observers to annotate large image collections much faster than before. We evaluated the method with three marine image datasets featuring different types of background, imaging equipment and object classes. Using MAIA, we were able to annotate objects of interest with an average recall of 84.1% more than twice as fast as compared to "traditional" annotation methods, which are purely based on software-supported direct visual inspection and manual annotation. The speed gain increases proportionally with the size of a dataset. The MAIA approach represents a substantial improvement on the path to greater efficiency in the annotation of large benthic image collections.

BTBR ob/ob mouse model of type 2 diabetes exhibits early loss of retinal function and retinal inflammation followed by late vascular changes.

AIMS/HYPOTHESIS: Diabetic retinopathy is increasing in prevalence worldwide and is fast becoming a global epidemic and a leading cause of visual loss. Current therapies are limited, and the development of effective treatments for diabetic retinopathy requires a greater in-depth knowledge of disease progression and suitable modelling of diabetic retinopathy in animals. The aim of this study was to assess the early pathological changes in retinal morphology and neuronal, inflammatory and vascular features consistent with diabetic retinopathy in the ob/ob mouse model of type 2 diabetes, to investigate whether features similar to those in human diabetic retinopathy were present. METHODS: Male and female wild-type (+/+), heterozygous (+/-) and homozygous (-/-) BTBR ob/ob mice were examined at 6, 10, 15 and 20 weeks of age. Animals were weighed and blood glucose was measured. TUNEL and brain-specific homeobox/POU domain protein 3A (BRN3A) markers were used to examine retinal ganglion cells. We used immunostaining (collagen IV and platelet endothelial cell adhesion molecule [PECAM]/CD31) to reveal retinal vessel degeneration. Spectral domain optical coherence tomography was used to reveal changes in the thickness and structure of the retinal layer. Vitreous fluorophotometry was used to investigate vascular permeability. A-waves, b-waves and oscillatory potentials were measured under photopic and scotopic conditions. Concanavalin A leucostasis and immunostaining with glial fibrillary acidic protein (GFAP) and ionised calcium-binding adapter molecule 1 (IBA-1) identified differences in inflammatory status. Paraffin sections and transmission electron microscopy were used to reveal changes in the thickness and structure of the retinal layer. RESULTS: Following the development of obesity and hyperglycaemia in 2-week-old and 3-week-old ob-/ob- mice, respectively (p < 0.001), early functional deficits (p < 0.001) and thinning of the inner retina (p < 0.001) were identified. Glial activation, leucostasis (p < 0.05) and a shift in microglia/macrophage phenotype were observed before microvascular degeneration (p < 0.05) and elevated vascular permeability occurred (p < 0.05). CONCLUSIONS/INTERPRETATION: The present characterisation of the development of diabetic retinopathy in the ob/ob mouse represents a platform that will enable the development of new therapies, particularly for the early stages of disease.

Spontaneous CNV in a novel mutant mouse is associated with early VEGF-A-driven angiogenesis and late-stage focal edema, neural cell loss, and dysfunction.

PURPOSE: Characterization of a mouse model of spontaneous choroidal neovascularization (sCNV) and its effect on retinal architecture and function. METHODS: The sCNV mouse phenotype was characterized by using fundus photography, fluorescein angiography, confocal scanning laser ophthalmoscopy (SLO), optical coherence tomography (OCT), ERG, immunostaining, biochemistry, and electron microscopy. A role for VEGF-A signaling in sCNV was investigated by using neutralizing antibodies and a role for macrophages explored by cell-depletion studies. RESULTS: The sCNV starts between postnatal day 10 and 15 (P10-P15), increasing in number and severity causing RPE disruption and dysfunction. Various morphological methods confirmed the choroidal origin and subretinal position of the angiogenic vessels. At approximately P25, vessels were present in the outer retina with instances of anastomosis of some sCNV lesions with the retinal vasculature. The number of CNV lesions was significantly decreased by systemic blockade of the VEGF-A pathway. Choroidal neovascularization size also was significantly modulated by reducing the number of lesion-associated macrophages. Later stages of sCNV were associated with edema, neuronal loss, and dysfunction. CONCLUSIONS: The sCNV mouse is a new model for the study of both early and late events associated with choroidal neovascularization. Pharmacological reduction in sCNV with VEGF-A antagonists and an anti-inflammatory strategy suggests the model may be useful for investigating novel targets for treating human ocular neovascular disease.

Sox7 and Sox17 are strain-specific modifiers of the lymphangiogenic defects caused by Sox18 dysfunction in mice.

Developmental defects caused by targeted gene inactivation in mice are commonly subject to strain-specific modifiers that modulate the severity of the phenotype. Although several genetic modifier loci have been mapped in mice, the gene(s) residing at these loci are mostly unidentified, and the molecular mechanisms of modifier action remain poorly understood. Mutations in Sox18 cause a variable phenotype in the human congenital syndrome hypotrichosis-lymphedema-telangiectasia, and the phenotype of Sox18-null mice varies from essentially normal to completely devoid of lymphatic vasculature and lethal, depending on the strain of the mice, suggesting a crucial role for strain-specific modifiers in this system. Here we show that two closely related Group F Sox factors, SOX7 and SOX17, are able to functionally substitute for SOX18 in vitro and in vivo. SOX7 and SOX17 are not normally expressed during lymphatic development, excluding a conventional redundancy mechanism. Instead, these genes are activated specifically in the absence of SOX18 function, and only in certain strains. Our studies identify Sox7 and Sox17 as modifiers of the Sox18 mutant phenotype, and reveal their mechanism of action as a novel mode of strain-specific compensatory upregulation.

Ex vivo magnetofection: a novel strategy for the study of gene function in mouse organogenesis.

Gene function during mouse development is often studied through the production and analysis of transgenic and knockout models. However, these techniques are time- and resource-consuming, and require specialized equipment and expertise. We have established a new protocol for functional studies that combines organ culture of explanted fetal tissues with microinjection and magnetically induced transfection ("magnetofection") of gene expression constructs. As proof-of-principle, we magnetofected cDNA constructs into genital ridge tissue as a means of gain-of-function analysis, and shRNA constructs for loss-of-function analysis. Ectopic expression of Sry induced female-to-male sex-reversal, whereas knockdown of Sox9 expression caused male-to-female sex-reversal, consistent with the known functions of these genes. Furthermore, ectopic expression of Tmem184a, a gene of unknown function, in female genital ridges, resulted in failure of gonocytes to enter meiosis. This technique will likely be applicable to the study of gene function in a broader range of developing organs and tissues.

Sox18 induces development of the lymphatic vasculature in mice.

The lymphatic system plays a key role in tissue fluid regulation and tumour metastasis, and lymphatic defects underlie many pathological states including lymphoedema, lymphangiectasia, lymphangioma and lymphatic dysplasia. However, the origins of the lymphatic system in the embryo, and the mechanisms that direct growth of the network of lymphatic vessels, remain unclear. Lymphatic vessels are thought to arise from endothelial precursor cells budding from the cardinal vein under the influence of the lymphatic hallmark gene Prox1 (prospero homeobox 1; ref. 4). Defects in the transcription factor gene SOX18 (SRY (sex determining region Y) box 18) cause lymphatic dysfunction in the human syndrome hypotrichosis-lymphoedema-telangiectasia, suggesting that Sox18 may also play a role in lymphatic development or function. Here we use molecular, cellular and genetic assays in mice to show that Sox18 acts as a molecular switch to induce differentiation of lymphatic endothelial cells. Sox18 is expressed in a subset of cardinal vein cells that later co-express Prox1 and migrate to form lymphatic vessels. Sox18 directly activates Prox1 transcription by binding to its proximal promoter. Overexpression of Sox18 in blood vascular endothelial cells induces them to express Prox1 and other lymphatic endothelial markers, while Sox18-null embryos show a complete blockade of lymphatic endothelial cell differentiation from the cardinal vein. Our findings demonstrate a critical role for Sox18 in developmental lymphangiogenesis, and suggest new avenues to investigate for therapeutic management of human lymphangiopathies.

Lymphatic vasculature: a molecular perspective.

The lymphatic vasculature comprises an intricate network of vessels critical for fluid homeostasis, immune surveillance and fat absorption. Recent studies have provided insights into the developmental processes and molecular mechanisms controlling the formation and remodelling of the lymphatic vessels. These studies have further demonstrated the essential and active role of the lymphatic vessels in various pathological conditions and advanced our understanding of the progression of human diseases, such as inflammation and tumorigenesis. In the context of the latest exciting findings, we review here the current understanding of the mechanisms of lymphatic development and contribution of lymphatic vessels to pathological conditions.

The VCAM-1 gene that encodes the vascular cell adhesion molecule is a target of the Sry-related high mobility group box gene, Sox18.

VCAM-1 (vascular cell adhesion molecule-1) and Sox18 are involved in vascular development. VCAM-1 is an important adhesion molecule that is expressed on endothelial cells and has a critical role in endothelial activation, inflammation, lymphatic pathophysiology, and atherogenesis. The Sry-related high mobility group box factor Sox18 has previously been implicated in endothelial pathologies. Mutations in human and mouse Sox18 leads to hypotrichosis and lymphedema. Furthermore, both Sox18 and VCAM-1 have very similar spatio-temporal patterns of expression, which is suggestive of cross-talk. We use biochemical techniques, cell culture systems, and the ragged opossum (RaOP) mouse model with a naturally occurring mutation in Sox18 to demonstrate that VCAM-1 is an important target of Sox18. Transfection, site-specific mutagenesis, and gel shift analyses demonstrated that Sox18 directly targeted and trans-activated VCAM-1 expression. Importantly, the naturally occurring Sox18 mutant attenuates the expression and activation of VCAM-1 in vitro. Furthermore, in vivo quantitation of VCAM-1 mRNA levels in wild type and RaOP mice demonstrates that RaOP animals show a dramatic and significant reduction in VCAM-1 mRNA expression in lung, skin, and skeletal muscle. Our observation that the VCAM-1 gene is an important target of SOX18 provides the first molecular insights into the vascular abnormalities in the mouse mutant ragged and the human hypotrichosis-lymphedema-telangiectasia disorder.

Sox18 mutations in the ragged mouse alleles ragged-like and opossum.

The ragged (Ra) spontaneous mouse mutant is characterised by abnormalities in its coat and cardiovascular system. Four alleles are known and we have previously described mutations in the transcription factor gene Sox18 in the Ra and Ra(J) alleles. We report here Sox18 mutations in the remaining two ragged alleles, opossum (Ra(op)) and ragged-like (Ragl). The single-base deletions cause a C-terminal frameshift, abolishing transcriptional trans-activation and impairing interaction with the partner protein MEF2C. The nature of these mutations, together with the near-normal phenotype of Sox18-null mice, suggests that the ragged mutant SOX18 proteins act in a dominant-negative fashion. The four ragged mutants represent an allelic series that reveal SOX18 structure-function relationships and implicate related SOX proteins in cardiovascular and hair follicle development.