Abstract
Deep learning solutions are being increasingly used in mobile applications. Although there are many open-source software tools for the development of deep learning solutions, there are no guidelines in one place in a unified manner for using these tools toward real-time deployment of these solutions on smartphones. From the variety of available deep learning tools, the most suited ones are used in this paper to enable real-time deployment of deep learning inference networks on smartphones. A uniform flow of implementation is devised for both Android and iOS smartphones. The advantage of using multi-threading to achieve or improve real-time throughputs is also showcased. A benchmarking framework consisting of accuracy, CPU/GPU consumption, and real-time throughput is considered for validation purposes. The developed deployment approach allows deep learning models to be turned into real-time smartphone apps with ease based on publicly available deep learning and smartphone software tools. This approach is applied to six popular or representative convolutional neural network models, and the validation results based on the benchmarking metrics are reported.
Highlights
Deep learning has had a dramatic impact on advancing the field of machine learning [1]. It has pushed the state of the art beyond what conventional approaches have achieved in various applications such as object detection [2], object localization [3], and speech recognition [4]
The expansion in the use of deep learning has been fueled by increases in the computational power of processors, in particular graphics processing units (GPUs), and the availability of large datasets for training
Deep learning involves deep neural networks (DNNs) consisting of a cascade of non-linear processing units arranged in layers
Summary
Deep learning has had a dramatic impact on advancing the field of machine learning [1]. Smartphones constitute the highest users of deep learning-based solutions spanning various applications such as voice assistants, automatic text prediction, and augmented reality They are used as research platforms to run deep learning solutions involving different applications such as concussion detection [8], jaundice diagnosis [9], schizophrenia recognition [10], and voice activity detection for hearing studies [11], among others. This work enables smartphones to be used as a portable research platform for deep learning studies Toward this objective, the rest of the paper is organized as follows: Section 2 describes the most suited deep learning libraries for smartphone deployment at the time of this writing, deployment steps based on the smartphone operating system, the software tools used to build deep learning apps, and the smartphone devices used to showcase a number of representative deep learning models.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
