scholarly journals A Novel Methodology for Measuring the Abstraction Capabilities of Image Recognition Algorithms

2021 ◽  
Vol 7 (8) ◽  
pp. 152
Author(s):  
Márton Gyula Hudáky ◽  
Péter Lehotay-Kéry ◽  
Attila Kiss
Keyword(s):  
Neural Network ◽  
Image Recognition ◽  
Intelligent Systems ◽  
Recognition Algorithms ◽  
Learning As A Process ◽  
Simple Neural Network ◽  
Recognition Software

Creating a widely excepted model on the measure of intelligence became inevitable due to the existence of an abundance of different intelligent systems. Measuring intelligence would provide feedback for the developers and ultimately lead us to create better artificial systems. In the present paper, we show a solution where learning as a process is examined, aiming to detect pre-written solutions and separate them from the knowledge acquired by the system. In our approach, we examine image recognition software by executing different transformations on objects and detect if the software was resilient to it. A system with the required intelligence is supposed to become resilient to the transformation after experiencing it several times. The method is successfully tested on a simple neural network, which is not able to learn most of the transformations examined. The method can be applied to any image recognition software to test its abstraction capabilities.

scholarly journals Investigation of integer neural networks for low cost embedded systems

2021 ◽  
Author(s):  
Thomas Behan
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Intelligent Systems ◽  
Low Cost ◽  
Cost Savings ◽  
Test Results ◽  
Trade Offs ◽  
Execution Speed ◽  
Multiple Levels ◽  
Simple Neural Network

Neural networks have been a topic of study for almost half a century and have become one of the predominant methods used for intelligent systems. During this time much progress has been made on improving the accuracy and expanding the capabilities of neural networks. This thesis is an investigation in a different direction, that is to reduce the computational requirements of neural networks to make them more suitable for implementation on very low end microcontrollers and DSPs. The goal is to understand the trade offs in cost, accuracy, and execution time on these low cost processors. To do this, two tests are performed. The first compares execution speed of a simple neural network on low cost hardware. This test demonstrates the advantages to using integer neural networks, and DSP operations. The second test, is a contract of the accuracy of an integer neural network and a floating-point neural network. This test uses a real world example and allows for testing multiple levels of quantization. The test results show the effects of quantization due to the use of integers, and show that there is a strong case for using integer neural networks on low cost microcontrolers, and that significant cost savings can be achieved in exchange for a very small reduction accuracy.

scholarly journals Investigation of integer neural networks for low cost embedded systems

2021 ◽  
Author(s):  
Thomas Behan
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Intelligent Systems ◽  
Low Cost ◽  
Cost Savings ◽  
Test Results ◽  
Trade Offs ◽  
Execution Speed ◽  
Multiple Levels ◽  
Simple Neural Network

Neural networks have been a topic of study for almost half a century and have become one of the predominant methods used for intelligent systems. During this time much progress has been made on improving the accuracy and expanding the capabilities of neural networks. This thesis is an investigation in a different direction, that is to reduce the computational requirements of neural networks to make them more suitable for implementation on very low end microcontrollers and DSPs. The goal is to understand the trade offs in cost, accuracy, and execution time on these low cost processors. To do this, two tests are performed. The first compares execution speed of a simple neural network on low cost hardware. This test demonstrates the advantages to using integer neural networks, and DSP operations. The second test, is a contract of the accuracy of an integer neural network and a floating-point neural network. This test uses a real world example and allows for testing multiple levels of quantization. The test results show the effects of quantization due to the use of integers, and show that there is a strong case for using integer neural networks on low cost microcontrolers, and that significant cost savings can be achieved in exchange for a very small reduction accuracy.

A SIMPLE NEURAL NETWORK APPROACH TO INVARIANT IMAGE RECOGNITION

2001 ◽  
Vol 15 (01) ◽  
pp. 11-17
Author(s):  
M. ANDRECUT ◽  
M. K. ALI
Keyword(s):  
Neural Network ◽  
Distribution Function ◽  
Image Recognition ◽  
Network Architecture ◽  
Cellular Neural Network ◽  
Angle Distribution ◽  
Neural Network Architecture ◽  
Feed Forward Neural Network ◽  
Neural Network Approach ◽  
Simple Neural Network

In this paper we propose a simple neural network architecture for invariant image recognition. The proposed neural network architecture contains three specialized modules. The neurons from the first module are connected in a cellular neural network structure, which is responsible for image processing: edge detection and segmentation. The second module is a feed forward neural network for invariant feature extraction from the sensorial layer: computation of the pair distribution function and bond angle distribution function. The third module is responsible for image classification. An application to the face recognition problem is also presented.

NEURAL NETWORK FOR IMAGE RECOGNITION OF MAGNETOOPTIC MATERIALS DOMAIN STRUCTURES

2016 ◽  
Vol 59 (3) ◽  
pp. 117-121
Author(s):  
A. V. Bragin ◽  
◽  
R. R. Navletov ◽  
D. V. Pyanzin ◽  
◽  
...  
Keyword(s):  
Neural Network ◽  
Image Recognition ◽  
Domain Structures

Unmanned aerial vehicle reconnaissance image recognition based on convolutional neural network

2021 ◽  
pp. 1-10
Author(s):  
Lipeng Si ◽  
Baolong Liu ◽  
Yanfang Fu
Keyword(s):  
Neural Network ◽  
Real Time ◽  
Image Recognition ◽  
Error Detection ◽  
Convolution Neural Network ◽  
Data Sets ◽  
Small Target ◽  
Network Algorithm ◽  
Neural Network Algorithm ◽  
Real Time Tracking

The important strategic position of military UAVs and the wide application of civil UAVs in many fields, they all mark the arrival of the era of unmanned aerial vehicles. At present, in the field of image research, recognition and real-time tracking of specific objects in images has been a technology that many scholars continue to study in depth and need to be further tackled. Image recognition and real-time tracking technology has been widely used in UAV aerial photography. Through the analysis of convolution neural network algorithm and the comparison of image recognition technology, the convolution neural network algorithm is improved to improve the image recognition effect. In this paper, a target detection technique based on improved Faster R-CNN is proposed. The algorithm model is implemented and the classification accuracy is improved through Faster R-CNN network optimization. Aiming at the problem of small target error detection and scale difference in aerial data sets, this paper designs the network structure of RPN and the optimization scheme of related algorithms. The structure of Faster R-CNN is adjusted by improving the embedding of CNN and OHEM algorithm, the accuracy of small target and multitarget detection is improved as a whole. The experimental results show that: compared with LENET-5, the recognition accuracy of the proposed algorithm is significantly improved. And with the increase of the number of samples, the accuracy of this algorithm is 98.9%.

Sign in / Sign up

Export Citation Format

Share Document