In this work, we investigate the use of deep learning for distortion-generic blind image quality assessment. We report on different design choices, ranging from the use of features extracted from pre-trained convolutional neural networks (CNNs) as a generic image description, to the use of features extracted from a CNN fine-tuned for the image quality task. Our best proposal, named DeepBIQ, estimates the image quality by average-pooling the scores predicted on multiple subregions of the original image. Experimental results on the LIVE In the Wild Image Quality Challenge Database show that DeepBIQ outperforms the state-of-the-art methods compared, having a linear correlation coefficient with human subjective scores of almost 0.91. These results are further confirmed also on four benchmark databases of synthetically distorted images: LIVE, CSIQ, TID2008, and TID2013.

On the use of deep learning for blind image quality assessment

Fabric defect detection systems and methods—A systematic literature review

DOI: 10.1002/admt.201900837 high-cost, and unsuitable for interaction with humans as well as intricate environments.[1–3] To overcome these challenges, great efforts have been taken to develop soft robots that are primarily made of compliant elastomers and polymers (silicone rubber, dielectric elastomers, liquidcrystalline elastomers and hydrogels, etc.) and possess unique properties such as lightweight, mechanical compliance, infinite degrees of freedom, continuous deformation, low cost, and easy fabrication.[2–6] Among these soft robotics, the class of soft crawling robots inspired from biological creatures has attracted increasing attention owing to their anticipated effective interaction with humans and uncertain environments, as well as potential capabilities of completing a variety of tasks, like search and rescue, infrastructure inspection, surveillance, drug delivery, and human assistance.[2,5] Various novel actuations and locomotion designs have been explored to drive soft crawling robots, with some of them even equipped with capabilities of the autonomous motion,[7] environment accommodation[4,8] and decision making.[9,10] For instance, soft crawling robots inspired by octopuses, worms, and jellyfishes have been designed to achieve complex motions with multiple gaits at low cost.[10,11] While there have been several excellent review papers on the topic of soft robotics,[3,4,12,13] there has not been a review paper that covers recent advances in soft crawling robots in depth. This review paper aims to fill that void. The locomotion mode, crawling speed, and working efficiency of soft crawling robots are mainly determined by the soft actuators they employed.[14] Generally, the external stimuli drive the actuators to generate the desired strains and/or deformations, and the induced strain and/or deformation will then supply the necessary propulsion for the robots to crawl. Up to now, a few actuation approaches have been successfully employed to drive the soft crawling robots, mainly consisting of pneumatic/hydraulic pressure,[15] chemical reaction,[16,18] and stimuli responses of soft active materials, such as dielectric elastomers (DEs),[19] shape memory alloys (SMAs),[20] magnetoactive elastomers (MAEs),[21] liquid-crystalline elastomers (LCEs),[22] piezoelectric materials (PEMs),[23] ionic polymer– metal composites (IPMCs),[24] and twisted and coiled polymers (TCPs).[25] Despite the inherent drawbacks of these actuation approaches, continuous efforts have been devoted to designing soft crawling robots capable of performing vivid, multigait, effective, and intelligent locomotion. Over the past decade, Soft crawling robots have attracted great attention due to their anticipated effective interactions with humans and uncertain environments, as well as their potential capabilities of completing a variety of tasks encompassing search and rescue, infrastructure inspection, surveillance, drug delivery, and human assistance. Herein, a comprehensive survey on recent advances of soft crawling robots categorized by their major actuation mechanisms is provided, including pneumatic/hydraulic pressure, chemical reaction, and soft active material-based actuations, which include dielectric elastomers, shape memory alloys, magnetoactive elastomers, liquid crystalline elastomers, piezoelectric materials, ionic polymer–metal composites, and twisted and coiled polymers. For each type of actuation, the prevalent modes of locomotion adopted in representative robots, the design, working principle and performance of their soft actuators, and the performance of each locomotion approach, as well as the advantages and drawbacks of each design are discussed. This review summarizes the state-of-the-art progresses and the critical knowledge in designing soft crawling robots and offers a guidance and insightful outlook for the future development of soft robots.

Soft Crawling Robots: Design, Actuation, and Locomotion

Digital image colorimetry on smartphone for chemical analysis: A review

Rapid and reliable identification of insects is important in many contexts, from the detection of disease vectors and invasive species to the sorting of material from biodiversity inventories. Beca ...

/pdf/automated-taxonomic-identification-of-insects-with-expert-34kapksv60.pdf

Automated Taxonomic Identification of Insects with Expert-Level Accuracy Using Effective Feature Transfer from Convolutional Networks

An auto fabric defect detection system via computer vision is used to replace manual inspection. In this paper, we propose a hardware accelerated algorithm based on a small-scale over-completed dictionary (SSOCD) via sparse coding (SC) method, which is realized on a parallel hardware platform (TMS320C6678). In order to reduce computation, the image patches projections in the training SSOCD are taken as features and the proposed features are more robust, and exhibit obvious advantages in detection results and computational cost. Furthermore, we introduce detection ratio and false ratio in order to measure the performance and reliability of the hardware accelerated algorithm. The experiments show that the proposed algorithm can run with high parallel efficiency and that the detection speed meets the real-time requirements of industrial inspection.

https://journals.sagepub.com/doi/pdf/10.5772/62058

Real-Time Fabric Defect Detection Using Accelerated Small-Scale Over-Completed Dictionary of Sparse Coding:

The aim of research on the no-reference image quality assessment problem is to design models that can predict the quality of distorted images consistently with human visual perception. Due to the little prior knowledge of the images, it is still a difficult problem. This paper proposes a computational algorithm based on hybrid model to automatically extract vision perception features from raw image patches. Convolutional neural network (CNN) and support vector regression (SVR) are combined for this purpose. In the hybrid model, the CNN is trained as an efficient feature extractor, and the SVR performs as the regression operator. Extensive experiments demonstrate very competitive quality prediction performance of the proposed method.

No-reference image quality assessment based on hybrid model

Blind image quality assessment in multiple bandpass and redundancy domains

Fine-grained image recognition, a computer vision task filled with challenges due to its imperceptible inter-class variance and large intra-class variance, has been drawing increasing attention. While manual annotation can be utilized to effectively enhance performance in this task, it is extremely time-consuming and expensive. Recently, Convolutional Neural Networks (CNN) achieved state-of-the-art performance in image classification. We propose a fine-grained image recognition framework by exploiting CNN as the raw feature extractor along with several effective methods including a feature encoding method, a feature weighting method, and a strategy to better incorporate information from multi-scale images to further improve recognition ability. Besides, we investigate two dimension reduction methods and successfully merge them to our framework to compact the final image representation. Based on the discriminative and compact framework, we achieved the state-of-the-art performance in terms of classification accuracy on several fine-grained image recognition benchmarks based on weekly supervision.

/pdf/refining-deep-convolutional-features-for-improving-fine-562falaksk.pdf

Refining deep convolutional features for improving fine-grained image recognition

In recent years, the foreign fibers in cotton lint significantly affect the quality of the final cotton textile products. It remains a challenging task to accurately distinguish foreign fibers from...

https://journals.sagepub.com/doi/pdf/10.1177/1729881419867600

Wenxuan Shi

Papers

Real-Time Fabric Defect Detection Using Accelerated Small-Scale Over-Completed Dictionary of Sparse Coding:

No-reference image quality assessment based on hybrid model

Blind image quality assessment in multiple bandpass and redundancy domains

Refining deep convolutional features for improving fine-grained image recognition

Classification of foreign fibers using deep learning and its implementation on embedded system