A Comparison of South East Asian Face Emotion Classification Based on Optimized Ellipse Data Using Clustering Technique

Muthukaruppan, Kartigayan; Thirugnanam, S.; Nagarajan, R.; Rizon, M.; Yaacob, Sazali; Muthukumaran, M.; Ramachandran, T.

doi:10.18178/joig.3.1.1-5

Cited by 20 publications

(3 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Impedance and near-infrared spectroscopy's effectiveness at determining out rough rice's moisture content was examined in one study. The findings showed the exceptional precision of both approaches, with near-infrared spectroscopy having a significantly higher accuracy (0.55% error) than impedance (0.79% error) [15], [16].…”

Section: Flir Thermal Studio Reportmentioning

confidence: 96%

Analysis of Rough Rice Moisture Content through Infrared Thermography (IRT)

Bejarin,

Fajardo

2023

IJMRA

View full text Add to dashboard Cite

Rough rice's moisture content should be measured in order to ensure its quality, preservation, and proper handling all through storage and processing. When assessing and monitoring the quality of agricultural products, such as rough rice, moisture content is an important consideration. For measuring the moisture content of various agricultural products, infrared thermography (IRT) is a quick and easy non-destructive technique that has shown effectiveness. Traditional methods for determining moisture content require a long time and require sample destruction. Images of the sample were captured with a FLIR ONE thermal camera. The average value of the pixels in the thermal image was applied to create a thermal index that represented each sample. This study aims to investigate the feasibility and accuracy of IRT as a method for analysing the moisture content of rough rice. The results of the study showed that 87.30% accuracy when measuring moisture content was achieved with infrared thermography. Based to the results, rough rice's moisture content may be accurately identified through infrared thermography evaluation using wavelengths analysis. The use of non-contact, non- destructive methods to measure the moisture content of rough rice can be highly beneficial to farmers, rice millers, and other relevant governmental organizations. The results demonstrate the potential of infrared thermography as a rapid and non-destructive tool for moisture content evaluation in rough rice, offering significant advantages in terms of time efficiency and sample preservation.

show abstract

Section: Flir Thermal Studio Reportmentioning

confidence: 96%

Analysis of Rough Rice Moisture Content through Infrared Thermography (IRT)

Bejarin,

Fajardo

2023

IJMRA

View full text Add to dashboard Cite

show abstract

“…Image processing is a component of the system's lower-level image analysis or computer vision functionality. With the aid of deep learning (DL), fuzzy systems, and improved feature extraction procedures, etc., image processing has made significant advances in recent years [1][2][3][4]. Currently, transformer models are gaining popularity in computer vision tasks.…”

Section: Introductionmentioning

confidence: 99%

How certain are tansformers in image classification: uncertainty analysis with Monte Carlo dropout

Islam

Zabeen

Islam

et al. 2023

Fifteenth International Conference on Machine Vision (ICMV 2022)

View full text Add to dashboard Cite

Researchers have been inspired to implement transformer models in solving machine vision problems after their tremendous success with natural language tasks. Using the straightforward architecture and swift performance of transformers, a variety of computer vision problems can be solved with more ease and effectiveness. However, a comparative evaluation of their uncertainty in prediction has not been done yet. As we know, real world applications require a measure of uncertainty to produce accurate predictions, which allows researchers to handle uncertain inputs and special cases, in order to successfully prevent overfitting. Our study approaches the unexplored issue of uncertainty estimation among three popular and effective transformer models employed in computer vision, such as Vision Transformers (ViT), Swin Transformers (SWT), and Compact Convolutional Transformers (CCT). We conduct a comparative experiment to determine which particular architecture is the most reliable in image classification. We use dropouts at the inference phase in order to measure the uncertainty of these transformer models. This approach, commonly known as Monte Carlo Dropout (MCD), works well as a low-complexity estimation to compute uncertainty. The MCD-based CCT model is the least uncertain architecture in this classification task. Our proposed MCD-infused CCT model also yields the best results with 78.4% accuracy, while the SWT model with embedded MCD exhibits Pmaximum performance gain where the accuracy increased by almost 3% with the final result being 71.4%.

show abstract

“…The category and orientation of each target were simultaneously estimated by finding the max of these classification scores. The original RICNN only conducted experiments on MNIST and fashion MNIST datasets [18][19][20][21] , which have small image size, large sample number and balanced categories. However, underwater target sonar images usually have large image size, small sample number and unbalanced categories.…”

Section: Introductionmentioning

confidence: 99%

Rotated target recognition of sonar images via convolutional neural networks with rotated inputs

Zhang

Tang²,

Zhong³

et al. 2022

Fourteenth International Conference on Digital Image Processing (ICDIP 2022)

View full text Add to dashboard Cite

Rotated target recognition is a challenge for Convolutional Neural Networks (CNN), and the current solution is to make CNN rotational invariant through data augmentation. However, data augmentation makes CNN easy to overfit small scale sonar image datasets, and increases its numbers of parameters and training time. This paper proposes to recognize rotated targets of sonar images using a novel CNN with Rotated Inputs (RICNN), which doesn’t need data augmentation. During training, RICNN was trained with sonar images of targets only at one orientation, which avoid it to learn multiple rotated versions of the same targets, and reduces both number of parameters and training time of CNN. During testing, RICNN calculated classification scores for each test image and its all-possible rotated versions. The max of these classification scores were used to simultaneously estimate the category and orientation of each target. Besides, to improve the generalization of RICNN on imbalanced sonar datasets, this paper also designs an imbalanced data sampler. Experiments on a self-made small, imbalanced sonar image rotated target recognition dataset show that the improved RICNN achieves 4.25% higher classification accuracy than data augmentation, and reduces the number of parameters and training time to 2.25% and 19.2% of that of data augmentation method. Moreover, RICNN achieves comparable orientation estimation accuracy with a CNN orientation regressor trained with data augmentation. Codes, dataset are publicly available.

show abstract

A Comparison of South East Asian Face Emotion Classification Based on Optimized Ellipse Data Using Clustering Technique

Cited by 20 publications

References 10 publications

Analysis of Rough Rice Moisture Content through Infrared Thermography (IRT)

Analysis of Rough Rice Moisture Content through Infrared Thermography (IRT)

How certain are tansformers in image classification: uncertainty analysis with Monte Carlo dropout

Rotated target recognition of sonar images via convolutional neural networks with rotated inputs

Contact Info

Product

Resources

About