Slim MobileNet: An Enhanced Deep Convolutional Neural Network

Bouguezzi, Safa; Faiedh, Hassène; Souani, Chokri

doi:10.1109/ssd52085.2021.9429519

Cited by 14 publications

(6 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Once the log mel spectrogram is created, the speech features presented on the spectrogram is analyzed by the Slim MobileNet (2D-CNN) and long short-term memory (LSTM) blocks. Slim MobileNet improves MobileNet-V1 model by reducing both the file size and the number of layers but with improved accuracy [15].…”

Section: Methodsmentioning

confidence: 99%

Artificial Intelligent-Based Wake Word Detection at Edge Device

Shen,

Fadhlullah,

Kamarulazizi

et al. 2023

JEVA

View full text Add to dashboard Cite

Deep Neural Network based wake word (such as Hi Alexa or Hey Siri) systems allow increasingly accurate speech communication between humans and machines. However, this setup requires high processing power or cloud services which may not be accessible by edge devices. Currently, the accuracy of machine learning methods for cloudless edge devices in voice activation hovers below 90%. This paper explores wake word implementation on edge devices using a 2-Dimensional Convolutional Neural Network (CNN) with improved and balanced accuracy and latency. The proposed CNN model is created, trained and quantized using TensorFlow on a PC and exported to a Raspberry Pi Zero 2 W. The quantization method reduces the model size by 20% and spectral gating is adopted to lower wake word inaccuracy detection in moderately noisy environment. The proposed system achieved more than 90% wake word detection accuracy across 30 to 50 dB background noise with an average of 1.03 second of response time for the intended user. The result shows low-powered edge device still offers competitive performance for detecting wake word without cloud services.

show abstract

Section: Methodsmentioning

confidence: 99%

Artificial Intelligent-Based Wake Word Detection at Edge Device

Shen,

Fadhlullah,

Kamarulazizi

et al. 2023

JEVA

View full text Add to dashboard Cite

show abstract

“…This streamlined model is currently applicable to a wide range of mobile devices and embedded visual platforms. Additionally, Thin MobileNet and Slim MobileNet models were proposed separately by Sinha et al [24] and Bouguezzi [25]. The size of the Thin MobileNet model is 9.9 MB, achieving an accuracy of 85.61%.…”

Section: Related Workmentioning

confidence: 99%

Comparison and Analysis of Three MobileNet-Based Models for Wildfire Detection

Du,

Li,

Noto

2024

JAIT

View full text Add to dashboard Cite

The dynamic equilibrium of ecosystems can be maintained through controlled burning, but excessive wildfires can lead to severe consequences. Therefore, the use of Internet of Things (IoT) devices equipped with deep image processing models for wildfire detection has recently become a trend. Conventional deep image processing models suffer from accuracy issues and large model sizes, limiting their applicability on small IoT devices. To address this challenge, we utilized lightweight deep image processing models such as the MobileNet series to train a wildfire database. Furthermore, we evaluated three different versions of MobileNet (V2, V3 Large, and V3 Small) using a crossentropy loss function to compare their accuracy and training times. Through data analysis, recommendations for deploying MobileNet models on IoT devices are provided. The results indicate that the ranking of MobileNet's accuracy from highest to lowest is V2, V3 Large, and V3 Small; the ranking of loss values from lowest to highest is V2, V3 Large, and V3 Small; and the ranking of training times from fastest to slowest is V3 Large, V2, and V3 Small.

show abstract

“…In this project, several classic CNNs models called ResNet50, MobileNet, and DenseNet169 was utilized to test their performance based on the FER-2013 dataset [4]. MobileNet model is a CNNs architecture that was constructed on an embedded board [10]. MobileNet was designed for efficient image classification on mobile and embedded devices, it was a pre-trained model with relatively fewer parameters.…”

Section: Cnn-based Emotion Recognition Modelmentioning

confidence: 99%

Investigation of multiple convolutional neural network models on emotion detection

Gao

2023

ACE

View full text Add to dashboard Cite

The accurate detection of emotions holds significant importance in the field of psychology, necessitating the careful selection of an appropriate model for facial expression classification. In this study, emotion detection is the classification task to compare the performance of MobileNet, ResNet, and DenseNet. For the detailed model, MobileNet, ResNet50, and DenseNet169 are selected for comparative analysis. The dataset FER-2013 is from Kaggle, which contains a training set and test set consisting of 29, 709 samples and 3589 samples, respectively, with seven facial expression categories. In terms of preprocessing, normalization, and data augmentation are considered. The whole dataset is normalized by dividing 255 and augmented with a Keras image generator. In the model-building step, the structure of the test models is controlled in the same structure. The pre-trained model from the Keras application connects with one global average pooling layer and adds one dense layer at the last as the output layer with the SoftMax activation function. Moreover, this study kept all hyper all parameters the same during the training period. After the model training, the confusion matrix is used to show the class relativity and the loss and accuracy of each model are plotted for analysis. Experimental results demonstrated that the MobileNet achieves 56.08% accuracy on test set which is more competitive than the DenseNet169 and ResNet50 and provides a relatively stable loss.

show abstract

Slim MobileNet: An Enhanced Deep Convolutional Neural Network

Cited by 14 publications

References 8 publications

Artificial Intelligent-Based Wake Word Detection at Edge Device

Artificial Intelligent-Based Wake Word Detection at Edge Device

Comparison and Analysis of Three MobileNet-Based Models for Wildfire Detection

Investigation of multiple convolutional neural network models on emotion detection

Contact Info

Product

Resources

About