scholarly journals Scale Ambiguities in Material Recognition

2021 ◽  
Author(s):  
Jacob Raleigh Cheeseman ◽  
Roland Fleming ◽  
Filipp Schmidt
Keyword(s):  
Spatial Scale ◽  
Apparent Distance ◽  
Image Features ◽  
Familiar Size ◽  
Top Down ◽  
Natural Materials ◽  
Material Recognition ◽  
Multi Scale ◽  
Verbal Instructions ◽  
Scale Structures

Many natural materials have complex, multi-scale structures. Consequently, the apparent identity of a surface can vary with the assumed spatial scale of the scene: a plowed field seen from afar can resemble corduroy seen up close. We investigated this ‘material-scale ambiguity’ using 87 photographs of diverse materials (e.g., water, sand, stone, metal, wood). Across two experiments, separate groups of participants (N = 72 adults) provided judgements of the material depicted in each image, either with or without manipulations of apparent distance (by verbal instructions, or adding objects of familiar size). Our results demonstrate that these manipulations can cause identical images to appear to belong to completely different material categories, depending on the perceived scale. Under challenging conditions, therefore, the perception of materials is susceptible to simple manipulations of apparent distance, revealing a striking example of top-down effects in the interpretation of image features.

scholarly journals Structure-Aware Residual Pyramid Network for Monocular Depth Estimation

2019 ◽  
Author(s):  
Xiaotian Chen ◽  
Xuejin Chen ◽  
Zheng-Jun Zha
Keyword(s):  
Local Structure ◽  
Feature Fusion ◽  
Depth Estimation ◽  
Image Features ◽  
Underlying Structure ◽  
Complex Scene ◽  
Multi Scale ◽  
Depth Prediction ◽  
Scale Structures ◽  
Monocular Depth

Monocular depth estimation is an essential task for scene understanding. The underlying structure of objects and stuff in a complex scene is critical to recovering accurate and visually-pleasing depth maps. Global structure conveys scene layouts, while local structure reflects shape details. Recently developed approaches based on convolutional neural networks (CNNs) significantly improve the performance of depth estimation. However, few of them take into account multi-scale structures in complex scenes. In this paper, we propose a Structure-Aware Residual Pyramid Network (SARPN) to exploit multi-scale structures for accurate depth prediction. We propose a Residual Pyramid Decoder (RPD) which expresses global scene structure in upper levels to represent layouts, and local structure in lower levels to present shape details. At each level, we propose Residual Refinement Modules (RRM) that predict residual maps to progressively add finer structures on the coarser structure predicted at the upper level. In order to fully exploit multi-scale image features, an Adaptive Dense Feature Fusion (ADFF) module, which adaptively fuses effective features from all scales for inferring structures of each scale, is introduced. Experiment results on the challenging NYU-Depth v2 dataset demonstrate that our proposed approach achieves state-of-the-art performance in both qualitative and quantitative evaluation. The code is available at https://github.com/Xt-Chen/SARPN.

Apparent distance as a function of familiar size.

1969 ◽  
Vol 79 (1, Pt.1) ◽  
pp. 109-115 ◽  
Author(s):  
Hiroshi Ono
Keyword(s):  
Apparent Distance ◽  
Familiar Size

scholarly journals A new multi-scale backbone network for object detection based on asymmetric convolutions

2021 ◽  
Vol 104 (2) ◽  
pp. 003685042110113
Author(s):  
Xianghua Ma ◽  
Zhenkun Yang
Keyword(s):  
Object Detection ◽  
Image Features ◽  
Detection Accuracy ◽  
Mobile Platforms ◽  
Multi Scale ◽  
Backbone Network ◽  
Aspect Ratios ◽  
Pascal Voc ◽  
Scale Characteristics ◽  
Detection Speed

Real-time object detection on mobile platforms is a crucial but challenging computer vision task. However, it is widely recognized that although the lightweight object detectors have a high detection speed, the detection accuracy is relatively low. In order to improve detecting accuracy, it is beneficial to extract complete multi-scale image features in visual cognitive tasks. Asymmetric convolutions have a useful quality, that is, they have different aspect ratios, which can be used to exact image features of objects, especially objects with multi-scale characteristics. In this paper, we exploit three different asymmetric convolutions in parallel and propose a new multi-scale asymmetric convolution unit, namely MAC block to enhance multi-scale representation ability of CNNs. In addition, MAC block can adaptively merge the features with different scales by allocating learnable weighted parameters to three different asymmetric convolution branches. The proposed MAC blocks can be inserted into the state-of-the-art backbone such as ResNet-50 to form a new multi-scale backbone network of object detectors. To evaluate the performance of MAC block, we conduct experiments on CIFAR-100, PASCAL VOC 2007, PASCAL VOC 2012 and MS COCO 2014 datasets. Experimental results show that the detection precision can be greatly improved while a fast detection speed is guaranteed as well.

scholarly journals Principles of seed banks and the emergence of complexity from dormancy

2021 ◽  
Vol 12 (1) ◽  
Author(s):  
Jay T. Lennon ◽  
Frank den Hollander ◽  
Maite Wilke-Berenguer ◽  
Jochen Blath
Keyword(s):  
Global Change ◽  
Metabolic Activity ◽  
Cancer Biology ◽  
Life Sciences ◽  
Seed Banks ◽  
Tree Of Life ◽  
Multi Scale ◽  
Emergent Phenomena ◽  
Scale Structures ◽  
Emergence Of Complexity

AbstractAcross the tree of life, populations have evolved the capacity to contend with suboptimal conditions by engaging in dormancy, whereby individuals enter a reversible state of reduced metabolic activity. The resulting seed banks are complex, storing information and imparting memory that gives rise to multi-scale structures and networks spanning collections of cells to entire ecosystems. We outline the fundamental attributes and emergent phenomena associated with dormancy and seed banks, with the vision for a unifying and mathematically based framework that can address problems in the life sciences, ranging from global change to cancer biology.

Metallography and Biomimetics – Or New Surfaces Without Chemistry?

2021 ◽  
Vol 58 (7) ◽  
pp. 446-459
Author(s):  
T. Fox ◽  
S. M. Lößlein ◽  
D. W. Müller ◽  
F. Mücklich
Keyword(s):  
Economic Cost ◽  
Electrical Contacts ◽  
Structured Surfaces ◽  
Multi Scale ◽  
Direct Laser ◽  
One Step ◽  
Fine Structures ◽  
Scale Structures ◽  
Processing Techniques ◽  
Direct Laser Interference Patterning

Abstract Fingerprints, a butterfly’s wings, or a lotus leaf: when it comes to surfaces, there is no such thing as coincidence in animated nature. Based on their surfaces, animals and plants control their wettability, their swimming resistance, their appearance, and much more. Evolution has optimized these surfaces and developed a microstructure that fits every need. It is all the more astonishing that, with regard to technical surfaces, man confines himself to random roughnesses or “smooth” surfaces. It is surely not a problem of a lack of incentives: structured surfaces have already provided evidence of optimizing friction and wear [1, 2, 3, 4], improving electrical contacts [5, 6], making implants biocompatible [7, 8], keeping away harmful bacteria [9], and much more. How come we continue counting on grinding, polishing, sandblasting, or etching? As so often, the problem can be found in economic cost effectiveness. It is possible to produce interesting structures such as those of the feather in Fig. 1. However, generating fine structures in the micro and nanometer range usually requires precise processing techniques. This is complex, time-consuming, and cannot readily be integrated into a manufacturing process. Things are different with Direct Laser Interference Patterning, DLIP) [10, 11]. This method makes use of the strong interference pattern of overlapped laser beams as a “stamp” to provide an entire surface area with dots, lines, or other patterns – in one shot. It thus saves time, allows for patterning speeds of up to 1 m2/min and does it without an elaborate pre- or post-treatment [10, 12]. The following article intends to outline how the method works, which structures can be generated, and how the complex multi-scale structures that nature developed over millions of years can be replicated in only one step.

scholarly journals Image Deblurring Using Multi-Stream Bottom-Top-Bottom Attention Network and Global Information-Based Fusion and Reconstruction Network

Sensors ◽  
2020 ◽  
Vol 20 (13) ◽  
pp. 3724
Author(s):  
Quan Zhou ◽  
Mingyue Ding ◽  
Xuming Zhang
Keyword(s):  
Deep Learning ◽  
Spatial Information ◽  
Structural Similarity ◽  
Image Deblurring ◽  
Image Features ◽  
Human Vision ◽  
Global Information ◽  
Attention Network ◽  
Multi Scale ◽  
Blind Image Deblurring

Image deblurring has been a challenging ill-posed problem in computer vision. Gaussian blur is a common model for image and signal degradation. The deep learning-based deblurring methods have attracted much attention due to their advantages over the traditional methods relying on hand-designed features. However, the existing deep learning-based deblurring techniques still cannot perform well in restoring the fine details and reconstructing the sharp edges. To address this issue, we have designed an effective end-to-end deep learning-based non-blind image deblurring algorithm. In the proposed method, a multi-stream bottom-top-bottom attention network (MBANet) with the encoder-to-decoder structure is designed to integrate low-level cues and high-level semantic information, which can facilitate extracting image features more effectively and improve the computational efficiency of the network. Moreover, the MBANet adopts a coarse-to-fine multi-scale strategy to process the input images to improve image deblurring performance. Furthermore, the global information-based fusion and reconstruction network is proposed to fuse multi-scale output maps to improve the global spatial information and recurrently refine the output deblurred image. The experiments were done on the public GoPro dataset and the realistic and dynamic scenes (REDS) dataset to evaluate the effectiveness and robustness of the proposed method. The experimental results show that the proposed method generally outperforms some traditional deburring methods and deep learning-based state-of-the-art deblurring methods such as scale-recurrent network (SRN) and denoising prior driven deep neural network (DPDNN) in terms of such quantitative indexes as peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) and human vision.

Sign in / Sign up

Export Citation Format

Share Document