Light guide plates are essential optical components widely used in a diverse range of applications ranging from medical lighting fixtures to back-lit TV displays. An essential step in the manufacturing of light guide plates is the quality inspection of defects such as scratches, bright/dark spots, and impurities. This is mainly done in industry through manual visual inspection for plate pattern irregularities, which is time-consuming and prone to human error and thus act as a significant barrier to high-throughput production. Advances in deep learning-driven computer vision has led to the exploration of automated visual quality inspection of light guide plates to improve inspection consistency, accuracy, and efficiency. However, given the computational constraints and high-throughput nature of real-world manufacturing environments, the widespread adoption of deep learning-driven visual inspection systems for inspecting light guide plates in real-world manufacturing environments has been greatly limited due to high computational requirements and integration challenges of existing deep learning approaches in research literature. In this work, we introduce a fully-integrated, high-throughput, high-performance deep learning-driven workflow for light guide plate surface visual quality inspection (VQI) tailored for real-world manufacturing environments. To enable automated VQI on the edge computing within the fully-integrated VQI system, a highly compact deep anti-aliased attention condenser neural network (which we name Light-DefectNet) tailored specifically for light guide plate surface defect detection in resource-constrained scenarios was created via machine-driven design exploration with computational and “best-practices” constraints as well as L1 paired classification discrepancy loss. Experiments show that Light-DetectNet achieves a detection accuracy of ∼98.2% on the LGPSDD benchmark while having just 770K parameters (∼33× and ∼6.9× lower than ResNet-50 and EfficientNet-B0, respectively) and ∼93M FLOPs (∼88× and ∼8.4× lower than ResNet-50 and EfficientNet-B0, respectively) and ∼8.8× faster inference speed than EfficientNet-B0 on an embedded ARM processor. As such, the proposed deep learning-driven workflow, integrated with the aforementioned LightDefectNet neural network, is highly suited for high-throughput, high-performance light plate surface VQI within real-world manufacturing environments.

While deep neural networks have been achieving state-of-the-art performance across a wide variety of applications, their vulnerability to adversarial attacks limits their widespread deployment for safety-critical applications. Alongside other adversarial defense approaches being investigated, there has been a very recent interest in improving adversarial robustness in deep neural networks through the introduction of perturbations during the training process. However, such methods leverage fixed, pre-defined perturbations and require significant hyper-parameter tuning that makes them very difficult to leverage in a general fashion. In this study, we introduce Learn2Perturb, an end-to-end feature perturbation learning approach for improving the adversarial robustness of deep neural networks. More specifically, we introduce novel perturbation-injection modules that are incorporated at each layer to perturb the feature space and increase uncertainty in the network. This feature perturbation is performed at both the training and the inference stages. Furthermore, inspired by the Expectation-Maximization, an alternating back-propagation training algorithm is introduced to train the network and noise parameters consecutively. Experimental results on CIFAR-10 and CIFAR-100 datasets show that the proposed Learn2Perturb method can result in deep neural networks which are 4-7% more robust on l_inf FGSM and PDG adversarial attacks and significantly outperforms the state-of-the-art against l_2 C\&W attack and a wide range of well-known black-box attacks.

Vehicle Make and Model Recognition (VMMR) systems provide a fully automatic framework to recognize and classify different vehicle models. Several approaches have been proposed to address this challenge; however, they can perform in restricted conditions. Here, in this paper, we formulate the VMMR as a fine-grained classification problem and propose a new configurable on-road VMMR framework. We benefit from the unsupervised feature learning methods, and in more details, we employ Locality-constraint Linear Coding (LLC) method as a fast feature encoder for encoding the input SIFT features. The proposed method can perform in real environments of different conditions. This framework can recognize 50 models of vehicles and has the advantage to classify every other vehicle not belonging to one of the specified 50 classes as an unknown vehicle. The proposed VMMR framework can be configured to become faster or more accurate based on the application domain. The proposed approach is examined on two datasets, including Iranian on-road vehicle (IORV) dataset and CompuCar dataset. The IORV dataset contains images of 50 models of vehicles captured in real situations by traffic-cameras in different weather and lighting conditions. The experimental results show the advantage of the real-time configuration of the proposed framework over the state-of-the-art methods on the IORV datatset and comparable results on CompuCar dataset with 97.5% and 98.4% accuracies, respectively and acceptable running time.

The approximation of nonlinear kernels via linear feature maps has recently gained interest due to their applications in reducing the training and testing time of kernel-based learning algorithms. Current random projection methods avoid the curse of dimensionality by embedding the nonlinear feature space into a low dimensional Euclidean space to create nonlinear kernels. We introduce a Layered Random Projection (LaRP) framework, where we model the linear kernels and nonlinearity separately for increased training efficiency. The proposed LaRP framework was assessed using the MNIST hand-written digits database and the COIL-100 object database, and showed notable improvement in object classification performance relative to other state-of-the-art random projection methods.

A novel approach to spatio-temporal saliency detection in video is proposed. Saliency computation is considered as an optimization problem that maximizes the energy of a fully-connected graphical model based on spatio-temporal feature distinctiveness. Each pixel in a video is modeled by a node, and the spatio-temporal feature distinctiveness between pixels by edges connecting the nodes in the graph. The computational complexity is addressed by compressing the fully-connected graph into an abstracted, fully-connected graph with far fewer nodes, where each node in the new graph characterizes nodal groups. The saliency value of each pixel is then computed based on spatio-temporal feature distinctiveness and the energy representation of its nodal group given the constructed graphical model. Experimental results show that our approach outperforms existing approaches to spatio-temporal salient region detection.

In this letter, a novel approach to visual saliency detection via neural response divergence (NeRD) is proposed, where synaptic portions of deep neural networks, previously trained for complex object recognition, are leveraged to compute low-level cues that can be used to compute image region distinctiveness. Based on this concept, an efficient visual saliency detection framework is proposed using deep convolutional StochasticNets. Experimental results using complex scene saliency dataset and MSRA10k natural image datasets show that the proposed NeRD approach can achieve improved performance when compared to state-of-the-art image saliency approaches, while attaining low computational complexity necessary for near-real-time computer vision applications.

Diffusion weighted magnetic resonance imaging (DW-MR) is a powerful tool in imaging-based prostate cancer screening and detection. Endorectal coils are commonly used in DW-MR imaging to improve the signal-to-noise ratio (SNR) of the acquisition, at the expense of significant intensity inhomogeneities (bias field) that worsens as we move away from the endorectal coil. The presence of bias field can have a significant negative impact on the accuracy of different image analysis tasks, as well as prostate tumor localization, thus leading to increased inter- and intra-observer variability. Retrospective bias correction approaches are introduced as a more efficient way of bias correction compared to the prospective methods such that they correct for both of the scanner and anatomy-related bias fields in MR imaging. Previously proposed retrospective bias field correction methods suffer from undesired noise amplification that can reduce the quality of bias-corrected DW-MR image. Here, we propose a unified data reconstruction approach that enables joint compensation of bias field as well as data noise in DW-MR imaging. The proposed noise-compensated, bias-corrected (NCBC) data reconstruction method takes advantage of a novel stochastically fully connected joint conditional random field (SFC-JCRF) model to mitigate the effects of data noise and bias field in the reconstructed MR data. The proposed NCBC reconstruction method was tested on synthetic DW-MR data, physical DW-phantom as well as real DW-MR data all acquired using endorectal MR coil. Both qualitative and quantitative analysis illustrated that the proposed NCBC method can achieve improved image quality when compared to other tested bias correction methods. As such, the proposed NCBC method may have potential as a useful retrospective approach for improving the consistency of image interpretations.

The reconstruction of high dynamic range (HDR) images via conventional camera systems and low dynamic range (LDR) images is a growing field of research in image acquisition. The radiance map associated with the HDR image of a scene is typically computed using multiple images of the same scene captured at different exposures (i.e., bracketed LDR imzages). This approach, though inexpensive, is sensitive to noise under high camera ISO. Each bracketed image is associated with a different level of noise due to the change in exposure time, and the noise is further amplified when tone-mapping the HDR image for display. A new framework is proposed to address the associated noise in the context of random fields. The estimation of the HDR image from a set of LDR images is formulated as a stochastically fully connected conditional random field where the spatial information is incorporated to compute the HDR value in combination with the LDR image values. Experimental results show that the proposed framework compensated the non-stationary ISO noise while preserving the boundaries in the estimated HDR images.

Markov random fields (MRFs) and conditional random fields (CRFs) are influential tools in image modeling, particularly for applications such as image segmentation. Local MRFs and CRFs utilize local nodal interactions when modeling, leading to excessive smoothness on boundaries (i.e., the short-boundary bias problem). Recently, the concept of fully connected conditional random fields with stochastic cliques (SFCRF) was proposed to enable long-range nodal interactions while addressing the computational complexity associated with fully connected random fields. While SFCRF was shown to provide significant improvements in segmentation accuracy, there were still limitations with the preservation of fine structure boundaries. To address these limitations, we propose a new approach to stochastic clique formation for fully connected random fields (G-SFCRF) that is guided by the structural characteristics of different nodes within the random field. In particular, fine structures surrounding a node are modeled statistically by probability distributions, and stochastic cliques are formed by considering the statistical similarities between nodes within the random fields. Experimental results show that G-SFCRF outperforms existing fully connected CRF frameworks, SFCRF, and the principled deep random field framework for image segmentation.

We present an illumination-compensation method to enable fast and reliable background subtraction under sudden, local illumination changes in wide area surveillance videos. We use Probabilistic Illumination Range Modeling (PIRM) to model the conditional probability distribution of current frame intensity given background intensity. With this model, we can identify a continuous range of current frame intensities that map to the same background intensity, and scale all pixels within that range in the current frame appropriately to enable illumination-compensated background subtraction. Experimental results using a standard academic dataset as well as very challenging industry videos show that PIRM can achieve improvements in compensating for sudden, local illumination changes.

A promising, recently explored, alternative to ultra-high b-value diffusion weighted imaging (UHB-DWI) is apparent ultra-high b-value diffusion-weighted image reconstruction (AUHB-DWR), where a computational model is used to assist in the reconstruction of apparent DW images at ultra-high b -values. Firstly, we present a novel approach to AUHB-DWR that aims to improve image quality. We formulate the reconstruction of an apparent DW image as a hidden conditional random field (HCRF) in which tissue model diffusion parameters act as hidden states in this random field. The second contribution of this paper is a new generation of fully connected conditional random fields, called the hidden stochastically fully connected conditional random fields (HSFCRF) that allows for efficient inference with significantly reduced computational complexity via stochastic clique structures. The proposed AUHB-DWR algorithms, HCRF and HSFCRF, are evaluated quantitatively in nine different patient cases using Fisher's criteria, probability of error, and coefficient of variation metrics to validate its effectiveness for the purpose of improving intensity delineation between expert identified suspected cancerous and healthy tissue within the prostate gland. The proposed methods are also examined using a prostate phantom, where the apparent ultra-high b-value DW images reconstructed using the tested AUHB-DWR methods are compared with real captured UHB-DWI. The results illustrate that the proposed AUHB-DWR methods has improved reconstruction quality and improved intensity delineation compared with existing AUHB-DWR approaches.

Conditional random fields (CRFs) are one of the most powerful frameworks in image modeling. However practical CRFs typically have edges only between nearby nodes; using more interactions and expressive relations among nodes make these methods impractical for large-scale applications, due to the high computational complexity. Recent work has shown that fully connected CRFs can be tractable by defining specific potential functions. In this paper, we present a novel framework to tackle the computational complexity of a fully connected graph without requiring specific potential functions. Instead, inspired by random graph theory and sampling methods, we propose a new clique structure called stochastic cliques. The stochastically fully connected CRF (SFCRF) is a marriage between random graphs and random fields, benefiting from the advantages of fully connected graphs while maintaining computational tractability. The effectiveness of SFCRF was examined by binary image labeling of highly noisy images. The results show that the proposed framework outperforms an adjacency CRF and a CRF with a large neighborhood size.

In this paper, a hierarchical model-based approach to frame rate-up conversion is presented. Given a sequence of consecutive video frames, a Spatio-Temporal Conditional Random Field (ST-CRF) is trained to capture both the motion and shape characteristics of objects within consecutive frames. A hierarchical tree is then constructed via hierarchical segmentation that sub-divides frames into regions based on color intensity and regional velocity. A hierarchical sampling approach is then introduced to construct new intermediate frames between adjacent video frames, where estimated intermediate frames are constructed at each level of a hierarchical tree constructed such that the probability of the ST-CRF is maximized. Preliminary results using videos with different motion characteristics show that the proposed approach has potential for producing intermediate frames with high visual quality.

We present Temporal Conditional Random Fields, a probabilistic framework for modeling object motion. The state-of-the-art discriminative approach for tracking is known as dynamic conditional random fields. This method models an event based on spatial and temporal relation between pixels in an image sequence without any prediction. To facilitate such a powerful graphical model with prediction and come up with a CRF-based predictor, we propose a set of new temporal relations for object tracking, with feature functions such as optical flow (calculated among consequent frames) and line filed features. We validate our proposed method with real data sequences and will show that the TCRF prediction is nearly equivalent with result of template matching. Experimental results indicate that our TCRF can predict future state of any maneuvering target with nearly zero error during its constant motion. Not only the proposed TCRF has a simple and easy to implement structure, but also it outperforms the state-of-the-art predictors such as Kalman filter.