Comparison of CNNs and SVM for voice control wheelchair

Sharifuddin, Mohammad Shahrul Izham; Nordin, Sharifalillah; Ali, Azliza Mohd

doi:10.11591/ijai.v9.i3.pp387-393

Cited by 14 publications

(7 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Begitu juga Terdapat banyak penelitian yang telah dilakukan tentang ekstraksi fitur sinyal wicara untuk menggerakan kursi roda. Pada penelitian ini terdapat empat perintah untuk menggerakkan kursi roda yaitu empat jenis perintah suara yang harus dikenali yaitu go, left, right dan stop [14]. Penelitian menggunakan model simulator dan prototipe kontroler berbasis ANFIS bersama dengan sinyal sensor online yang dapat memaksimalkan kinerja kursi roda dan meningkatkan kualitas hidup penyandang disabilitas [15].…”

Section: Pendahuluanunclassified

“…Pada penelitian tersebut, terdapat penelitian untuk menggerakkan kursi roda dengan empat perintah [14]. Sehingga untuk penelitian yang dilakukan yaitu menggunakan lima perintah (sinyal wivaea yang digunakan ada lima perintah, yaitu: maju, mundur, kiri, kanan dan berhenti).…”

Section: Pendahuluanunclassified

“…Pada peneltian ini, untuk menggerakkan kursi roda dengan lima perintah (berhenti, maju, mundur, kiri, dan kanan) digunakan ekstrasi fitur yang berbeda dari peneliti sebelumnya [14], yaitu menggunakan metode FFT dikombinasikan dengan metode Wavelet. Terdapat satu nilai ekstrasi ciri yang diambil dari FFT yaitu nilai Ratarata Energi serta diambil dari dua nilai ekstrasi ciri metode wavelet yaitu Nilai subband energi dan Nilai Median of zero crossing threshold.…”

Section: Blok Diagram Sistemunclassified

See 2 more Smart Citations

Klasifikasi Sinyal Wicara Untuk Gerakan Kursi Roda Menggunakan Metode Jaringan Syaraf Tiruan Backpropagation

Wisaksono,

Hindarto,

Efiyanti

et al. 2024

Transmisi J.Ilm. Tek. Elekt.

View full text Add to dashboard Cite

Kursi Roda yang dikendalikan oleh sinyal wicara adalah teknologi bantuan yang inovatif dirancang untuk meningkatkan mobilitas dan kemandirian penyandang disabilitas. Penelitian ini bertujuan untuk mengkalsifikasi sinyal wicara yang digunakan untuk Gerakan kursi roda yang dapat dioperasikan menggunakan perintah suara. Ada lima jenis data perintah wicara yang harus dikenali yaitu maju, mundur, kiri, kanan, dan berhenti. Terdapat beberapa ekstrasi fitur yang digunakan, yaitu menggunakan metode FFT dengan mengambil nilai energi rata-rata, dan menggunakan metode DWT dengan mengambil nilai subband energi dan nilai zero crossing threshold. Proses klasifikasi yang digunakan menggunakan metode Jaringan Syaraf Tiruan Backpropagation. Penelitian yang diujikan menggunakan metode Jringn Syaraf Tiruan menghasilkan tingkat akurasi yang lebih baik dengan menggunakan hidden layer 2 dan hidden layer 3. Hasil akurasi yang didapatkan sebesar 100% untuk proses pelatihan dan 100% untuk proses pengujian

show abstract

Section: Pendahuluanunclassified

Section: Blok Diagram Sistemunclassified

See 1 more Smart Citation

Klasifikasi Sinyal Wicara Untuk Gerakan Kursi Roda Menggunakan Metode Jaringan Syaraf Tiruan Backpropagation

Wisaksono,

Hindarto,

Efiyanti

et al. 2024

Transmisi J.Ilm. Tek. Elekt.

View full text Add to dashboard Cite

show abstract

“…The Raspberry Pi 4 Model B was selected as the platform for deploying the speaker identification model, considering its available resources [22,23]. The Raspbian 64-bit operating system was installed first, followed by the libraries needed to run Python scripts.…”

Section: Embedded Deploymentmentioning

confidence: 99%

Siamese Networks for Speaker Identification on Resource-Constrained Platforms

Jie,

Zabidi,

Sadiah

et al. 2023

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

This paper investigates the implementation of a lightweight Siamese neural network for enhancing speaker identification accuracy and inference speed in embedded systems. Integrating speaker identification into embedded systems can improve portability and versatility. Siamese neural networks achieve speaker identification by comparing input voice samples to reference voices in a database, effectively extracting features and classifying speakers accurately. Considering the trade-off between accuracy and complexity, as well as hardware constraints in embedded systems, various neural networks could be applied to speaker identification. This paper compares the incorporation of CNN architectures targeted for embedded systems, MCUNet, SqueezeNet and MobileNetv2, to implement Siamese neural networks on a Raspberry Pi. Our experiments demonstrate that MCUNet achieves 85% accuracy with a 0.23-second inference time. In comparison, the larger MobileNetv2 attains 84.5% accuracy with a 0.32-second inference time. Additionally, contrastive loss was superior to binary cross-entropy loss in the Siamese neural network. The system using contrastive loss had almost 68% lower loss scores, resulting in a more stable performance and more accurate predictions. In conclusion, this paper establishes that an appropriate lightweight Siamese neural network, combined with contrastive loss, can significantly improve speaker identification accuracy, and enable efficient deployment on resource-constrained platforms.

show abstract

“…Convolutional neural network is widely used for image classification [4]. It has had ground-breaking results over the past decade from image recognition to voice recognition [5][6] [7]. Convolutional Neural Network were used in projects such as diagnosis of cancer using histopathological images and traffic sign classification [8].…”

Section: Convolutional Neural Networkmentioning

confidence: 99%

Plant Recognition System Using Convolutional Neural Network

Zefri

Nordin

2022

IOP Conf. Ser.: Earth Environ. Sci.

Self Cite

View full text Add to dashboard Cite

There are many plant species available to man, even some that have not been discovered. There are species of plants are reported to be on the verge of extinction. Due to the inhumane nature of humans, forest is cleared to make space for industrial purposes. These purposes will lead to the destruction of nature, including plants. Plants are important to be preserved for the future. Plants should be made aware to the public to avoid such disaster. Ensuring they know the plants species are the reasons for the development of this system. The system will ensure people can identify plants without having flipping through research papers or books. Those plants that are similar to one another can be recognized easily. Using convolutional neural network to recognize plant can help the public gain more awareness towards plants species. Using the traditional convolutional neural network for the training of model to recognize plants. The system was tested to get the best accuracy to recognize plants. The images were increased to ensure a better accuracy. This system requires a good model in order to recognize the correct plants. Convolutional neural network is commonly used for image classification due to its high accuracy. The accuracy for the system created in this project is 78.85%. This project can be improved in the future by increasing the accuracy. There are a lot of ways to increase the accuracy such as trying out other CNN architectures or by feeding the system more images. This project involves 7 phases which is preliminary study, knowledge gathering and acquisition, knowledge representation, system design, system development, system testing and evaluation, and documentation.

show abstract

Comparison of CNNs and SVM for voice control wheelchair

Cited by 14 publications

References 29 publications

Klasifikasi Sinyal Wicara Untuk Gerakan Kursi Roda Menggunakan Metode Jaringan Syaraf Tiruan Backpropagation

Klasifikasi Sinyal Wicara Untuk Gerakan Kursi Roda Menggunakan Metode Jaringan Syaraf Tiruan Backpropagation

Siamese Networks for Speaker Identification on Resource-Constrained Platforms

Plant Recognition System Using Convolutional Neural Network

Contact Info

Product

Resources

About