Fusion based En-FEC Transfer Learning Approach for Automobile Parts Recognition System

Prasad, Mukesh; Rajora, Shantanu; Gupta, Deepak; Daraghmi, Yousef‐Awwad; Daraghmi, Eman; Yadav, Pranay; Tiwari, Prayag; Saxena, Amit

doi:10.1109/ssci.2018.8628789

Cited by 7 publications

(3 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Prasad et al [30] introduce a novel deep learning method to classify the various parts of any running engine, such as rocker arms, crank shafts, distributors, assecory belts, air ducts, etc organized in the car.…”

Section: Related Work On Medical Image Interpolationmentioning

confidence: 99%

Enhancement of Corneal Images using High Resolution Techniques

El-Hameed

Al-Hanafy

Ashiba³

et al. 2023

Preprint

View full text Add to dashboard Cite

This paper presents an enhancement method to deal with the medical images which have low resolution as corneal images that have hexagonal nature and contain edges which must be preserved. So its resolution should be increased to help ophthalmologists to accurately diagnose and monitor diseases. Image interpolation is employed for resolution enhancement. In this paper, we consider interpolation techniques such as: polynomial interpolation, adaptive polynomial interpolation, inverse interpolation, and super-resolution (SR) reconstruction in order to enhance corneal image interpolation-based and learning-based techniques. The polynomial interpolation includes: bilinear, bicubic, cubic spline and cubic O-MOMS as well as their adaptive techniques. While the inverse interpolation techniques comprises linear minimum mean square error (LMMSE), maximum entropy, and a regularized image interpolation. Although polynomial based techniques are the most popular due to their simplicity, they don’t take into account the local activity levels of the image to be interpolated and cause a blurring effect in data. Interpolation is applied to each pixel to adapt for changing local activity levels, this adaptation reduces the blur effect and provides a better visual quality. Polynomial interpolation and its adaptive techniques are called signal synthesis techniques because they are based on the use of known neighborhoods to synthesize unknown pixel values. However, they did not meet the prerequisites of modern sampling theory in the interpolation process, so a pre-filtering step is required in the reconstruction process; a correction filter is needed before the interpolation process to compensate for the non-ideality of the image acquisition model. This correction filter is obtained as the inverse of the cross-correlation sequence between the acquisition model filter and the reconstruction filter, hence the inverse interpolation technique is applied which is superior where low image degradation model is taken into consideration. Finally the SR technique is proposed and compared to other previous techniques. Simulations are conducted to investigate the performance of the considered proposed techniques. It is shown that the adaptive polynomial image interpolation methods and inverse interpolation techniques give satisfactory results in terms of peak signal-to-noise ratio (PSNR), while the proposed SR technique is the most superior.

show abstract

Section: Related Work On Medical Image Interpolationmentioning

confidence: 99%

Enhancement of Corneal Images using High Resolution Techniques

El-Hameed

Al-Hanafy

Ashiba³

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…In industrial part classification, Hong et al introduced a method to relabel the detection result of vehicle parts [29]. Prasad et al proposed a novel deep learning approach to classify the various parts of any operational engine; it can qualitatively classify and henceforth give the corresponding class label of the machinery/engine part under consideration [30]. Abraham et al developed a two-level machine learning-based system to classify different car parts [31].…”

Section: Industrial Part Classificationmentioning

confidence: 99%

“…Prasad et al. proposed a novel deep learning approach to classify the various parts of any operational engine; it can qualitatively classify and henceforth give the corresponding class label of the machinery/engine part under consideration [30]. Abraham et al.…”

Section: Related Workmentioning

confidence: 99%

A DenseNet‐based feature weighting convolutional network recognition model and its application in industrial part classification

An,

Sun,

Song

et al. 2023

IET Image Processing

View full text Add to dashboard Cite

Traditional warehousing typically needs machine learning or manual tagging to classify objects. However, this method is less robust and consumes a lot of labour and material resources. Based on DenseNet, this work proposes a feature weighting convolutional network recognition model and designs a set of software and hardware for data acquisition, which is applied to the efficient classification of industrial parts in warehouse management. Firstly, this work modifies DenseNet by embedding SE‐Block, and replaces the cross‐entropy loss function with the focus loss function to optimize the model structure. Secondly, a multi‐view hardware and software acquisition system is designed to complete the functions of part image acquisition, image preprocessing, model training and part recognition. Finally, an industrial parts sorting experiment was designed. Compared with the original DenseNet model, the proposed weighted convolutional network identification model showed that the accuracy of the modified model was increased by 3.09% and the convergence rate was significantly improved. The modified model proposed in this work aims to improve the recognition accuracy of industrial parts in modern warehouse management, so as to modify the classification efficiency of warehouse parts in production.

show abstract