A comparative analysis of automatic deep neural networks for image retrieval

Al-Jubouri, Hanan; Mahmmod, Sawsan M.

doi:10.12928/telkomnika.v19i3.18157

Cited by 8 publications

(6 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Simultaneously, mean time represented an average measure of prediction time for each model, as detailed in Table III. In Tables II, MAcc for each DNN model is calculated as the simple average of recorded accuracy values across all images (N=10), using the formula (1) [50], [51]:…”

Section: Experimental Procedures 1) Baseline Modelmentioning

confidence: 99%

Baseline model for deep neural networks in resource-constrained environments: an empirical investigation

Careem,

Gapar Md Johar,

Khatibi

2024

IJCDS

View full text Add to dashboard Cite

This paper presents an empirical study on advanced Deep Neural Network (DNN) models, with a focus on identifying potential baseline models for efficient deployment in resource-constrained environments (RCE). The systematic evaluation encompasses ten state-of-the-art pre-trained DNN models: ResNet50, InceptionResNetV2, InceptionV3, MobileNet, MobileNetV2, EfficientNetB0, EfficientNetB1, EfficientNetB2, DenseNet121, and Xception, within the context of an RCE setting. Evaluation criteria, such as parameters (indicating model complexity), storage space (reflecting storage requirements), CPU usage time (for real-time applications), and accuracy (reflecting prediction truth), are considered through systematic experimental procedures. The results highlight MobileNet's excellent trade-off between accuracy and resource requirements, especially in terms of CPU and storage consumption, in experimental scenarios where image predictions are performed on an RCE device. Utilizing the identified baseline model, a new model, GRM-MobileNet, was developed by implementing compound scaling and global average pooling techniques. GRM-MobileNet exhibits a substantial reduction of 23.81% in parameters compared to MobileNet, leading to a model size that is 23.88% smaller. Moreover, GRM-MobileNet demonstrates a significant improvement in accuracy, achieving a remarkable gain of 28.12% over MobileNet. Although the enhancement in inference time for GRM-MobileNet compared to MobileNet is modest at 1.66%, the overall improvements underscore the effectiveness of the employed strategies in enhancing the model's performance. A future study will examine other model optimization strategies, including factorization and pruning, which ultimately lead to faster inference without compromising accuracy, in an effort to improve the efficiency of the GRM-MobileNet model and its inference time.

show abstract

Section: Experimental Procedures 1) Baseline Modelmentioning

confidence: 99%

Baseline model for deep neural networks in resource-constrained environments: an empirical investigation

Careem,

Gapar Md Johar,

Khatibi

2024

IJCDS

View full text Add to dashboard Cite

show abstract

“…Experiments show that the suggested method works better than the current situation. [5] Al-Jubouri, H. A., & Mahmmod, S. M. [6] A lot of the work that CBIR does for picture recognition depends on feature descriptions to bridge the language gap and pull out visual characteristics. We used known precision measures to compare different methods to ones that had been tried before.…”

Section: Related Workmentioning

confidence: 99%

Analyzing Retrieval Method Using Classification and Clustering Process of Remote Sensor Satellite Images

Kumari D.,

2024

IJMCR

View full text Add to dashboard Cite

Today, remote sensors and satellite images are utilized across enormous regions. As visual tools, they are very important in cities, agriculture, disaster relief, and tracking systems. Pattern recognition, classification, and clustering are examples of machine learning techniques that can be used to acquire and retrieve the images. This is because they can accurately categorize and classify the images into clusters, which will help search engines discover the image that was given as a question. In our study, we look at Naïve Bayes, SVM Linear and Non-Linear, and Random Forest as classification methods, and Agglomerative Hierarchy and BIRCH as clustering methods. In this study, the fast recovery method and high accuracy rate of image retrieval by classifiers from machine learning were compared with clustering techniques from transfer learning. Using the UC Merced dataset, this research demonstrates that the accuracy classifier as SVM linear is 64.6%, SVM non-linear is 77.7%, random forest is 66.8%, and naïve bayes is 60.3%, and clustering of Agglomerative is 96.5% and BIRCH is 92.8%. The work that was mentioned is one of a kind because the clustering methods work better than the classification methods when they are used. It's unsupervised learning whose scores are above 90% for all classes.

show abstract

“…Then, a ranked list of the most top 5 similar images is returned that may help a radiologist or specialist doctor to make the right decision. Since the similarity measurement is another issue in CBIR [31], [32] our experiments tested City-block (DF1) and cosine (DF2) distance functions as ( 6) and (7).…”

Section: Image Retrievalmentioning

confidence: 99%

Chest Radiographs Images Retrieval Using Deep Learning Networks

Mahmoud

Al-Jubouri²,

Abdoulabbas³

2022

Bulletin EEI

View full text Add to dashboard Cite

Chest diseases are among the most common diseases today. More than one million people with pneumonia enter the hospital, and about 50,000 people die annually in the U.S. alone. Also, Coronavirus disease (COVID-19) is a risky disease that threatens the health by affecting the lungs of many people around the world. Chest X-ray and CT-scan images are the radiological imaging that can be helpful to detect COVID-19. A radiologist would need to compare a patient's image with the most similar images. Content-based image retrieval in terms of medical images offers such a facility based on visual feature descriptor and similarity measurements. In this paper, a retrieval algorithm was developed to tackle such challenges based on deep convolutional neural networks (e.g., ResNet-50, AlexNet, and GoogleNet) to produce an effective feature descriptor. Also, similarity measures such as City block and Cosine were employed to compare two images. Chest X-ray and CT-scan datasets used to evaluate the proposed algorithms with a highest performance applying ResNet -50 (99% COVID-19 (+) and 98% COVID-19 (–)) and GoogleNet (84% COVID-19 (+) and 81% COVID-19 (–)) for X-ray and CT-scan respectively. The percentage increased about 1-4% when voting was used by a k-nearest neighbor classifier

show abstract

A comparative analysis of automatic deep neural networks for image retrieval

Cited by 8 publications

References 22 publications

Baseline model for deep neural networks in resource-constrained environments: an empirical investigation

Baseline model for deep neural networks in resource-constrained environments: an empirical investigation

Analyzing Retrieval Method Using Classification and Clustering Process of Remote Sensor Satellite Images

Chest Radiographs Images Retrieval Using Deep Learning Networks

Contact Info

Product

Resources

About