Autonomous Vehicle Dataset with Real Multi-Driver Scenes and Biometric Data.

The development of autonomous vehicles is becoming increasingly popular and gathering real-world data is considered a valuable task. Many datasets have been published recently in the autonomous vehicle sector, with synthetic datasets gaining particular interest due to availability and cost. For a real implementation and correct evaluation of vehicles at higher levels of autonomy, it is also necessary to consider human interaction, which is precisely something that lacks in existing datasets. In this article the UPCT dataset is presented, a public dataset containing high quality, multimodal data obtained using state-of-the-art sensors and equipment installed onboard the UPCT's CICar autonomous vehicle. The dataset includes data from a variety of perception sensors including 3D LiDAR, cameras, IMU, GPS, encoders, as well as driver biometric data and driver behaviour questionnaires. In addition to the dataset, the software developed for data synchronisation and processing has been made available. The quality of the dataset was validated using an end-to-end neural network model with multiple inputs to obtain the speed and steering wheel angle and it obtained very promising results.

A novel ground truth multispectral image dataset with weight, anthocyanins, and Brix index measures of grape berries tested for its utility in machine learning pipelines.

BACKGROUND: The combination of computer vision devices such as multispectral cameras coupled with artificial intelligence has provided a major leap forward in image-based analysis of biological processes. Supervised artificial intelligence algorithms require large ground truth image datasets for model training, which allows to validate or refute research hypotheses and to carry out comparisons between models. However, public datasets of images are scarce and ground truth images are surprisingly few considering the numbers required for training algorithms. RESULTS: We created a dataset of 1,283 multidimensional arrays, using berries from five different grape varieties. Each array has 37 images of wavelengths between 488.38 and 952.76 nm obtained from single berries. Coupled to each multispectral image, we added a dataset with measurements including, weight, anthocyanin content, and Brix index for each independent grape. Thus, the images have paired measures, creating a ground truth dataset. We tested the dataset with 2 neural network algorithms: multilayer perceptron (MLP) and 3-dimensional convolutional neural network (3D-CNN). A perfect (100% accuracy) classification model was fit with either the MLP or 3D-CNN algorithms. CONCLUSIONS: This is the first public dataset of grape ground truth multispectral images. Associated with each multispectral image, there are measures of the weight, anthocyanins, and Brix index. The dataset should be useful to develop deep learning algorithms for classification, dimensionality reduction, regression, and prediction analysis.

gcProfileMakeR: An R Package for Automatic Classification of Constitutive and Non-Constitutive Metabolites.

Metabolomes comprise constitutive and non-constitutive metabolites produced due to physiological, genetic or environmental effects. However, finding constitutive metabolites and non-constitutive metabolites in large datasets is technically challenging. We developed gcProfileMakeR, an R package using standard Excel output files from an Agilent Chemstation GC-MS for automatic data analysis using CAS numbers. gcProfileMakeR has two filters for data preprocessing removing contaminants and low-quality peaks. The first function NormalizeWithinFiles, samples assigning retention times to CAS. The second function NormalizeBetweenFiles, reaches a consensus between files where compounds in close retention times are grouped together. The third function getGroups, establishes what is considered as Constitutive Profile, Non-constitutive by Frequency i.e., not present in all samples and Non-constitutive by Quality. Results can be plotted with the plotGroup function. We used it to analyse floral scent emissions in four snapdragon genotypes. These included a wild type, Deficiens nicotianoides and compacta affecting floral identity and RNAi:AmLHY targeting a circadian clock gene. We identified differences in scent constitutive and non-constitutive profiles as well as in timing of emission. gcProfileMakeR is a very useful tool to define constitutive and non-constitutive scent profiles. It also allows to analyse genotypes and circadian datasets to identify differing metabolites.

The Snapdragon LATE ELONGATED HYPOCOTYL Plays A Dual Role in Activating Floral Growth and Scent Emission.

The plant circadian clock controls a large number of internal processes, including growth and metabolism. Scent emission displays a circadian pattern in many species such as the snapdragon. Here we show that knocking down LATE ELONGATED HYPOCOTYL in Antirrhinum majus affects growth and scent emission. In order to gain an understanding of the growth kinetics, we took a phenomic approach using in-house artificial vision systems, obtaining time-lapse videos. Wild type flowers showed a higher growth speed than knockdown plants. The maximal growth rate was decreased by 22% in plants with lower LHY expression. Floral volatiles were differentially affected as RNAi plants showed advanced emission of compounds synthesized from cinnamic acid and delayed emission of metabolites of benzoic acid. The monoterpenes myrcene and ocimene were delayed, whereas the sesquiterpene farnesene was advanced. Overall, transgenic lines showed an altered volatile emission pattern and displayed a modified scent profile. Our results show that AmLHY plays an important role in the quantitative and qualitative control of floral growth and scent emission.

The Petunia CHANEL Gene is a ZEITLUPE Ortholog Coordinating Growth and Scent Profiles.

The floral perianth, comprising sepals and petals, conceals the sexual organs and attracts pollinators. The coordination of growth and scent emission is not fully understood. We have analyzed the effect of knocking down CHANEL (PhCHL), the ZEITLUPE ortholog in petunia (PhCHL) by hairpin RNAs. Plants with low PhCHL mRNA had overall decreased size. Growth evaluation using time lapse image analysis showed that early leaf movement was not affected by RNAi:PhCHL, but flower angle movement was modified, moving earlier during the day in knockdown plants than in wild types. Despite differences in stem length, growth rate was not significantly affected by loss of PhCHL. In contrast, petal growth displayed lower growth rate in RNAi:PhCHL. Decreased levels of PhCHL caused strongly modified scent profiles, including changes in composition and timing of emission resulting in volatile profiles highly divergent from the wild type. Our results show a role of PhCHL in controlling growth and development of vegetative and reproductive organs in petunia. The different effects of PhCHL on organ development indicate an organ-specific interpretation of the down regulation of PhCHL. Through the control of both timing and quantitative volatile emissions, PhCHL appears to be a major coordinator of scent profiles.

A Systematic Review of Perception System and Simulators for Autonomous Vehicles Research.

This paper presents a systematic review of the perception systems and simulators for autonomous vehicles (AV). This work has been divided into three parts. In the first part, perception systems are categorized as environment perception systems and positioning estimation systems. The paper presents the physical fundamentals, principle functioning, and electromagnetic spectrum used to operate the most common sensors used in perception systems (ultrasonic, RADAR, LiDAR, cameras, IMU, GNSS, RTK, etc.). Furthermore, their strengths and weaknesses are shown, and the quantification of their features using spider charts will allow proper selection of different sensors depending on 11 features. In the second part, the main elements to be taken into account in the simulation of a perception system of an AV are presented. For this purpose, the paper describes simulators for model-based development, the main game engines that can be used for simulation, simulators from the robotics field, and lastly simulators used specifically for AV. Finally, the current state of regulations that are being applied in different countries around the world on issues concerning the implementation of autonomous vehicles is presented.

Plant phenomics: an overview of image acquisition technologies and image data analysis algorithms.

The study of phenomes or phenomics has been a central part of biology. The field of automatic phenotype acquisition technologies based on images has seen an important advance in the last years. As with other high-throughput technologies, it addresses a common set of problems, including data acquisition and analysis. In this review, we give an overview of the main systems developed to acquire images. We give an in-depth analysis of image processing with its major issues and the algorithms that are being used or emerging as useful to obtain data out of images in an automatic fashion.

A Machine Learning Approach to Pedestrian Detection for Autonomous Vehicles Using High-Definition 3D Range Data.

This article describes an automated sensor-based system to detect pedestrians in an autonomous vehicle application. Although the vehicle is equipped with a broad set of sensors, the article focuses on the processing of the information generated by a Velodyne HDL-64E LIDAR sensor. The cloud of points generated by the sensor (more than 1 million points per revolution) is processed to detect pedestrians, by selecting cubic shapes and applying machine vision and machine learning algorithms to the XY, XZ, and YZ projections of the points contained in the cube. The work relates an exhaustive analysis of the performance of three different machine learning algorithms: k-Nearest Neighbours (kNN), Naïve Bayes classifier (NBC), and Support Vector Machine (SVM). These algorithms have been trained with 1931 samples. The final performance of the method, measured a real traffic scenery, which contained 16 pedestrians and 469 samples of non-pedestrians, shows sensitivity (81.2%), accuracy (96.2%) and specificity (96.8%).

Defect Detection in Textures through the Use of Entropy as a Means for Automatically Selecting the Wavelet Decomposition Level.

This paper presents a robust method for defect detection in textures, entropy-based automatic selection of the wavelet decomposition level (EADL), based on a wavelet reconstruction scheme, for detecting defects in a wide variety of structural and statistical textures. Two main features are presented. One of the new features is an original use of the normalized absolute function value (NABS) calculated from the wavelet coefficients derived at various different decomposition levels in order to identify textures where the defect can be isolated by eliminating the texture pattern in the first decomposition level. The second is the use of Shannon's entropy, calculated over detail subimages, for automatic selection of the band for image reconstruction, which, unlike other techniques, such as those based on the co-occurrence matrix or on energy calculation, provides a lower decomposition level, thus avoiding excessive degradation of the image, allowing a more accurate defect segmentation. A metric analysis of the results of the proposed method with nine different thresholding algorithms determined that selecting the appropriate thresholding method is important to achieve optimum performance in defect detection. As a consequence, several different thresholding algorithms depending on the type of texture are proposed.

Machine Learning and Computer Vision System for Phenotype Data Acquisition and Analysis in Plants.

Phenomics is a technology-driven approach with promising future to obtain unbiased data of biological systems. Image acquisition is relatively simple. However data handling and analysis are not as developed compared to the sampling capacities. We present a system based on machine learning (ML) algorithms and computer vision intended to solve the automatic phenotype data analysis in plant material. We developed a growth-chamber able to accommodate species of various sizes. Night image acquisition requires near infrared lightning. For the ML process, we tested three different algorithms: k-nearest neighbour (kNN), Naive Bayes Classifier (NBC), and Support Vector Machine. Each ML algorithm was executed with different kernel functions and they were trained with raw data and two types of data normalisation. Different metrics were computed to determine the optimal configuration of the machine learning algorithms. We obtained a performance of 99.31% in kNN for RGB images and a 99.34% in SVM for NIR. Our results show that ML techniques can speed up phenomic data analysis. Furthermore, both RGB and NIR images can be segmented successfully but may require different ML algorithms for segmentation.

Automatic detection of microaneurysms in diabetic retinopathy fundus images using the L*a*b color space.

We develop an automated image processing system for detecting microaneurysm (MA) in diabetic patients. Diabetic retinopathy is one of the main causes of preventable blindness in working age diabetic people with the presence of an MA being one of the first signs. We transform the eye fundus images to the L*a*b* color space in order to separately process the L* and a* channels, looking for MAs in each of them. We then fuse the results, and last send the MA candidates to a k-nearest neighbors classifier for final assessment. The performance of the method, measured against 50 images with an ophthalmologist's hand-drawn ground-truth, shows high sensitivity (100%) and accuracy (84%), and running times around 10 s. This kind of automatic image processing application is important in order to reduce the burden on the public health system associated with the diagnosis of diabetic retinopathy given the high number of potential patients that need periodic screening.

Development of a configurable growth chamber with a computer vision system to study circadian rhythm in plants.

Plant development is the result of an endogenous morphogenetic program that integrates environmental signals. The so-called circadian clock is a set of genes that integrates environmental inputs into an internal pacing system that gates growth and other outputs. Study of circadian growth responses requires high sampling rates to detect changes in growth and avoid aliasing. We have developed a flexible configurable growth chamber comprising a computer vision system that allows sampling rates ranging between one image per 30 s to hours/days. The vision system has a controlled illumination system, which allows the user to set up different configurations. The illumination system used emits a combination of wavelengths ensuring the optimal growth of species under analysis. In order to obtain high contrast of captured images, the capture system is composed of two CCD cameras, for day and night periods. Depending on the sample type, a flexible image processing software calculates different parameters based on geometric calculations. As a proof of concept we tested the system in three different plant tissues, growth of petunia- and snapdragon (Antirrhinum majus) flowers and of cladodes from the cactus Opuntia ficus-indica. We found that petunia flowers grow at a steady pace and display a strong growth increase in the early morning, whereas Opuntia cladode growth turned out not to follow a circadian growth pattern under the growth conditions imposed. Furthermore we were able to identify a decoupling of increase in area and length indicating that two independent growth processes are responsible for the final size and shape of the cladode.

Alzheimer's disease amyloid peptides interact with DNA, as proved by surface plasmon resonance.

According to the amyloid hypothesis, abnormal processing of the ß-amyloid precursor protein in Alzheimer's disease patients increases the production of ß-amyloid toxic peptides, which, after forming highly aggregated fibrillar structures, lead to extracellular plaques formation, neuronal loss and dementia. However, a great deal of evidence has point to intracellular small oligomers of amyloid peptides, probably transient intermediates in the process of fibrillar structures formation, as the most toxic species. In order to study the amyloid-DNA interaction, we have selected here three different forms of the amyloid peptide: Aß1-40, Aß25-35 and a scrambled form of Aß25-35. Surface Plasmon Resonance was used together with UV-visible spectroscopy, Electrophoresis and Electronic Microscopy to carry out this study. Our results prove that, similarly to the full length Aß1-42, all conformations of toxic amyloid peptides, Aß1-40 and Aß25-35, may bind DNA. In contrast, the scrambled form of Aß25-35, a non-aggregating and nontoxic form of this peptide, could not bind DNA. We conclude that although the amyloid-DNA interaction is closely related to the amyloid aggregation proneness, this cannot be the only factor which determines the interaction, since small oligomers of amyloid peptides may also bind DNA if their predominant negatively charged amino acid residues are previously neutralized.

A novel method to increase LinLog CMOS sensors' performance in high dynamic range scenarios.

Images from high dynamic range (HDR) scenes must be obtained with minimum loss of information. For this purpose it is necessary to take full advantage of the quantification levels provided by the CCD/CMOS image sensor. LinLog CMOS sensors satisfy the above demand by offering an adjustable response curve that combines linear and logarithmic responses. This paper presents a novel method to quickly adjust the parameters that control the response curve of a LinLog CMOS image sensor. We propose to use an Adaptive Proportional-Integral-Derivative controller to adjust the exposure time of the sensor, together with control algorithms based on the saturation level and the entropy of the images. With this method the sensor's maximum dynamic range (120 dB) can be used to acquire good quality images from HDR scenes with fast, automatic adaptation to scene conditions. Adaptation to a new scene is rapid, with a sensor response adjustment of less than eight frames when working in real time video mode. At least 67% of the scene entropy can be retained with this method.