Image Classification Using Fusion of Multiple Neural Networks

Fan, Jiayi; Lee, JangHyeon; Lee, YongKeun

doi:10.1109/itc-cscc52171.2021.9501468

Cited by 2 publications

(5 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The computational complexity of the known neural network models [11][12][13][14][15][16][17][18] is one of their disadvantages. A solution to this problem can be found in the use of direct (Kronecker) penetrating product of matrices for the analytical description of operations performed in a concrete DCNN layer (expressions ( 10)-( 24)).…”

Section: Discussion Of the Results Obtained In The Study Of Recognizing The Object Images In Aerial Photographs Using Dcnnmentioning

confidence: 99%

“…Networks of direct data propagation [12][13][14][15] in which attribute values (the image) of the object being classified are fed to the input and a label (the class name) or a numeric class code is formed at the output are components of the DCNNs (CNNs) architecture most often used for classification. In the proposed model, an RGB image in JPEG format with a dimensionality of 227×227×3 is fed to the neural network input and a class label is attached at the output (Table 1).…”

Section: The Study Materials and Methodsmentioning

confidence: 99%

“…It was shown in [15] that the CNNs are widely used in applications including those for image classification. However, in the case of limitations and disadvantages of a single CNN, the image recognition accuracy can be further improved by using multiple neural networks.…”

Section: Literature Review and Problem Statementmentioning

confidence: 99%

“…where B 4 is a block matrix of pixels ₿ 4 prepared for the max pool operation in which similarly to (15), 27×27×256 blocks of 3×3 pixels in size are formed with a discrete step of 2 pixels; ₿ 5 is the output block matrix of the layer (Max pool 2).…”

Section: Second Maxpooling Layer (Max Pool 2)mentioning

confidence: 99%

“…The third maxpooling layer (Max pool 3) This layer is essentially identical to the two previous Max pool layers (15), (17) with the difference that the maxpooling layer processes in this case a block matrix of data with the dimensionality of 13×13×256 elements.…”

Section: The Fifth Convolutional Layer (Conv 5)mentioning

confidence: 99%

See 4 more Smart Citations

Improvement of the model of object recognition in aero photographs using deep convolutional neural networks

Slyusar

Protsenko

Chernukha

et al. 2021

EEJET

View full text Add to dashboard Cite

Detection and recognition of objects in images is the main problem to be solved by computer vision systems. As part of solving this problem, the model of object recognition in aerial photographs taken from unmanned aerial vehicles has been improved. A study of object recognition in aerial photographs using deep convolutional neural networks has been carried out. Analysis of possible implementations showed that the AlexNet 2012 model (Canada) trained on the ImageNet image set (China) is most suitable for this problem solution. This model was used as a basic one. The object recognition error for this model with the use of the ImageNet test set of images amounted to 15 %. To solve the problem of improving the effectiveness of object recognition in aerial photographs for 10 classes of images, the final fully connected layer was modified by rejection from 1,000 to 10 neurons and additional two-stage training of the resulting model. Additional training was carried out with a set of images prepared from aerial photographs at stage 1 and with a set of VisDrone 2021 (China) images at stage 2. Optimal training parameters were selected: speed (step) (0.0001), number of epochs (100). As a result, a new model under the proposed name of AlexVisDrone was obtained. The effectiveness of the proposed model was checked with a test set of 100 images for each class (the total number of classes was 10). Accuracy and sensitivity were chosen as the main indicators of the model effectiveness. As a result, an increase in recognition accuracy from 7 % (for images from aerial photographs) to 9 % (for the VisDrone 2021 set) was obtained which has indicated that the choice of neural network architecture and training parameters was correct. The use of the proposed model makes it possible to automate the process of object recognition in aerial photographs. In the future, it is advisable to use this model at ground stations of unmanned aerial vehicle complex control when processing aerial photographs taken from unmanned aerial vehicles, in robotic systems, in video surveillance complexes and when designing unmanned vehicle systems

show abstract