China 
Africa 
Explore chapters and articles related to this topic
Applying deep learning to the classification of exercise electrocardiography symptoms
Published in Artde Donald Kin-Tak Lam, Stephen D. Prior, Sheng-Joue Young, Siu-Tsen Shen, Liang-Wen Ji, System Innovation in a Post-Pandemic World, 2022
Chun-Yen Chen, Shie-Jue Lee, Hsiang-Chun Lee, Ching-Yi Tsa, Su-Te Chen, Yu-Ju Li
Our DNN model was developed using Keras (https://keras.io) and Tensor-Flow (https://www.tensorflow.org) whose libraries are written in Python. The experiments were carried out on a computer with Intel(R) Core(TM) i5-10400 CPU, 32GB RAM, and an NVIDIA GeForce RTX 2080 super graphics processing unit (GPU). We used the Stratified 5-Fold method for training and testing in our experiments. The stratified 5-fold method was used to conduct cross-validation.
Automatic detection and tracking of precast walls from surveillance construction site videos
Published in Airong Chen, Xin Ruan, Dan M. Frangopol, Life-Cycle Civil Engineering: Innovation, Theory and Practice, 2021
Z.C. Wang, B. Yang, Q.L. Zhang
In this section, experiments are conducted with implementations of Mask R-CNN and DeepSORT, which are forked from corresponding official open source projects on Github. The proposed framework is trained and tested with system environment of Windows 10, Python 3.6, Keras 2.1.6, TensorFlow 1.12.0, CUDA 9.0, and cuDNN 7.1.4 on a computer equipped with a Intel Core i7-8700k CPU @3.20GHz, 16 GB DDR4 memory, and 8GB memory GeForce RTX 2080 graphics processing unit (GPU).
Image classification of Philippine bird species using deep learning
Published in Shin-ya Nishizaki, Masayuki Numao, Jaime Caro, Merlin Teodosia Suarez, Theory and Practice of Computation, 2019
C.R. Raquel, K.M.A. Alarcon, L.L. Figueroa
The computational neural network (CNN) was trained and tested using the Philippine bird species dataset. The training time ran for approximately 24 hours. The network was run on a Hewlett-Packard Envy Notebook having an Intel Core i76500U CPU, 8GB memory with an NVIDIA GEFORCE GTX graphics processing unit (GPU). The model achieved an accuracy of 93% in the train set with a loss score of 0.54. The test set accuracy is 94%. The confusion matrix (see Appendix) shows that majority of the bird species in the test set have been classified correctly. For example, all the images of the bird specie Philippine Scops-Owl have been correctly classified (27 out of 27 images found on the 32nd row of the confusion matrix). This is possibly due to its unique features such as having a round red eye and unique head shape compared to other birds.
Traffic sign extraction using deep hierarchical feature learning and mobile light detection and ranging (LiDAR) data on rural highways
Published in Journal of Intelligent Transportation Systems, 2023
Maged Gouda, Alexander Epp, Rowan Tilroe, Karim El-Basyouny
The model was trained in 200 epochs, each epoch consisting of 845 batches of 24 cubes (sampled as described above). The Adam (Kingma & Ba, 2014) optimizer was used with a learning rate that starts at 0.001 and halves every 15 epochs. The loss function minimized during training was unweighted cross-entropy. The GPU used is a GeForce RTX 2080 Ti with 11 GB VRAM.
A PPM-based UNet for Tumour and Kidney Segmentation in CTScans
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2023
Marcus Vinicius S. L. Oliveira, Caio E. F. Matos, Geraldo Braz Júnior, Anselmo Cardoso de Paiva, João D. Sousa de Almeida, Gabriel J. S. Costa, Matheus L. L. Bessa, Mario P. Freitas Filho
In this section, we present the experiments to evaluate the proposed method. The method was built in python using the Keras library, with Tensorflow GPU backend. The GPU used for the experiments was the NVIDIA GeForce RTX 3060.