Light-weight shadow detection via GCN-based annotation strategy and knowledge distillation

An essential element for intelligent perception in mechatronic and robotic systems (M&RS) is the visual object detection algorithm. With the ever-increasing advance of artificial neural networks (ANN), researchers have proposed numerous ANN-based visual object detection methods that have proven to be effective. However, networks with cumbersome structures do not befit the real-time scenarios in M&RS, necessitating the techniques of model compression. In the paper, a novel approach to training light-weight visual object detection networks is developed by revisiting knowledge distillation. Traditional knowledge distillation methods are oriented towards image classification is not compatible with object detection. Therefore, a variant of knowledge distillation is developed and adapted to a state-of-the-art keypoint-based visual detection method. Two strategies named as positive sample retaining and early distribution softening are employed to yield a natural adaption. The mutual consistency between teacher model and student model is further promoted through a hint-based distillation. By extensive controlled experiments, the proposed method is testified to be effective in enhancing the light-weight network’s performance by a large margin.

Download Full-text

Ultrafast Video Attention Prediction with Coupled Knowledge Distillation

Proceedings of the AAAI Conference on Artificial Intelligence ◽

10.1609/aaai.v34i07.6710 ◽

2020 ◽

Vol 34 (07) ◽

pp. 10802-10809

Author(s):

Kui Fu ◽

Peipei Shi ◽

Yafei Song ◽

Shiming Ge ◽

Xiangju Lu ◽

...

Keyword(s):

State Of The Art ◽

Network Models ◽

Light Weight ◽

Low Resolution ◽

Neural Network Models ◽

Teacher Networks ◽

Memory Footprint ◽

Large Memory ◽

Knowledge Distillation ◽

Acceleration Method

Large convolutional neural network models have recently demonstrated impressive performance on video attention prediction. Conventionally, these models are with intensive computation and large memory. To address these issues, we design an extremely light-weight network with ultrafast speed, named UVA-Net. The network is constructed based on depth-wise convolutions and takes low-resolution images as input. However, this straight-forward acceleration method will decrease performance dramatically. To this end, we propose a coupled knowledge distillation strategy to augment and train the network effectively. With this strategy, the model can further automatically discover and emphasize implicit useful cues contained in the data. Both spatial and temporal knowledge learned by the high-resolution complex teacher networks also can be distilled and transferred into the proposed low-resolution light-weight spatiotemporal network. Experimental results show that the performance of our model is comparable to 11 state-of-the-art models in video attention prediction, while it costs only 0.68 MB memory footprint, runs about 10,106 FPS on GPU and 404 FPS on CPU, which is 206 times faster than previous models.

Download Full-text

Microstructure of an extruded rapidly solidified Al-8Fe-2Mo alloy

Proceedings, annual meeting, Electron Microscopy Society of America ◽

10.1017/s0424820100144218 ◽

1986 ◽

Vol 44 ◽

pp. 548-549

Author(s):

W. T. Donlon ◽

J. E. Allison ◽

S. Shinozaki

Keyword(s):

Electron Microscopy ◽

Automotive Industry ◽

Fuel Economy ◽

High Strength ◽

Al Alloys ◽

Light Weight ◽

Thin Sections ◽

Strength And Durability ◽

Rapidly Solidified ◽

Whitney Aircraft

Light weight materials which possess high strength and durability are being utilized by the automotive industry to increase fuel economy. Rapidly solidified (RS) Al alloys are currently being extensively studied for this purpose. In this investigation the microstructure of an extruded Al-8Fe-2Mo alloy, produced by Pratt & Whitney Aircraft, Goverment Products Div. was examined in a JE0L 2000FX AEM. Both electropolished thin sections, and extraction replicas were examined to characterize this material. The consolidation procedure for producing this material included a 9:1 extrusion at 340°C followed by a 16:1 extrusion at 400°C, utilizing RS powders which have also been characterized utilizing electron microscopy.

Download Full-text

Light-Weight Tire Concept

Tire Science and Technology ◽

10.2346/1.2137514 ◽

1996 ◽

Vol 24 (2) ◽

pp. 119-131

Author(s):

F. Lux ◽

H. Stumpf

Keyword(s):

Fuel Consumption ◽

Automobile Industry ◽

Rolling Resistance ◽

Performance Standards ◽

Light Weight ◽

Production Methods ◽

Lower Fuel Consumption ◽

Material Resources ◽

The Past ◽

Tire Production

Abstract Current demands by the consumer, the automobile industry, and the environment have determined the basis of this investigation. In the past, the requirements—ever faster, ever sportier—were accepted as decisive parameters for the development of our study. In the future, rational and safety-related tire characteristics as well as environmental consciousness will increase, whereas purely performance-related parameters will diminish in their importance. Through our light-weight tire project, we have paved the way for future tire generations. The first priority is the minimal use of material resources; this means a reduction of materials and energy in tire production by using advanced design and production methods without sacrificing performance standards. This benefits the consumer—the final judge of all of our activities—by considerably reducing the rolling resistance, leading to lower fuel consumption. Further design targets include the improvement of rolling behavior and increased comfort by reducing tire weight, and therefore a reduction in unsprung masses on the vehicle.

Download Full-text

Electrolytic analysis with platinum electrodes of light weight

American Journal of Science ◽

10.2475/ajs.s4-34.200.107 ◽

1912 ◽

Vol s4-34 (200) ◽

pp. 107-112

Author(s):

F. A. Gooch ◽

W. L. Burdick

Keyword(s):

Light Weight ◽

Platinum Electrodes

Download Full-text

Determinants Considered by Consumers for Brand Selection - with Reference to Laptops in Albania

Creative and Knowledge Society ◽

10.2478/cks-2014-0008 ◽

2014 ◽

Vol 4 (2) ◽

Author(s):

Margret Plloçi ◽

Macit Koc

Keyword(s):

Primary Data ◽

Battery Life ◽

Decision Making Process ◽

Light Weight ◽

Sources Of Information ◽

Technical Features ◽

The Relationship ◽

The City ◽

Different Sources ◽

Core Features

Abstract Purpose of the article There is relatively a big number of brands in the market of laptops nowadays in Albania. It appears that the number of brands offered in this market could easily be compared to the number of brands in Europe and even broader. The purpose of this study is to help Albanian vendors understand the criteria that consumers take into consideration when they make the decision to purchase a laptop. Methodology/methods The research is based on the collection and the analyses of the primary data collected through interviews to people like managers or employees who work in the sector of trading laptops or in businesses like education where laptops are broadly used recently; then a survey is done through a questionnaire delivered to customers who already own and use a laptop and customers who are potential buyers of laptops. Scientific aim The aim of the research is to identify if there are any relationships between the demographics of the consumers and the criteria of buying a laptop; on the other hand, to find out how is the relationship between the demographics and the features of different brands. Findings The study found out that Albanian consumers have good knowledge of laptops and their brands, and they use different sources of information for making their decisions in buying a laptop; it is found that there are relationships between some demographics like age or gender and the appraisal for some attributes of the laptops like price, design and high graphics card; it is also found that some technical features and other attributes of using laptops are some of the determinants that influence the laptops’ purchases. Conclusions It is realized that one of the most important demographics of the consumers is their age. Some core features like RAM, ROM, battery life, processor quality, light weight or attributes that are connected to the purposes of using the laptop computers like practicality and mobility in using them, work and studying processes, quick access to the internet are determinant factors which influence the decision making process of purchasing a laptop. I would recommend that future researches be focused also on the relationship between the customers’ income and their preferred brand or ranking brands according to the customers’ preferences. Such studies should also extend outside the city of Tirana.

Download Full-text