CNNdroid

Oskouei, Seyyed Salar Latifi; Golestani, Hossein Bakhshi; Hashemi, Matin; Ghiasi, Soheil

doi:10.1145/2964284.2973801

Cited by 72 publications

(6 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While machine learning frameworks implement different tools and methods to automatically optimize their configurations, many parameters still require manual tuning. For example, finding an optimal batch size to maximize 1 throughput with latency constraints requires direct measurements of the inference performance on the execution platform (see Figures 5,6,10,11,15,18). Manual placement of operations between GPU and CPU can also significantly improve the execution performance of an inference model (Section 5.1).…”

Section: Discussionmentioning

confidence: 99%

“…Recently Qualcomm announced hardware acceleration support for TensorFlow using their latest Snapdragon SoC [3]. Some research prototypes that leverage mobile device special purpose processors (e.g., DSP, GPU) also exist [13,[15][16][17][18]. Other recent research has looked at the computational behavior of CNNs and the impact of the neural network architecture, such as number of layers, depth, etc., on it [5,11,27].…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Latency and throughput characterization of convolutional neural networks for mobile computer vision

Hanhirova

Kämäräinen

Seppälä

et al. 2018

Proceedings of the 9th ACM Multimedia Systems Conference

View full text Add to dashboard Cite

We study performance characteristics of convolutional neural networks (CNN) for mobile computer vision systems. CNNs have proven to be a powerful and efficient approach to implement such systems. However, the system performance depends largely on the utilization of hardware accelerators, which are able to speed up the execution of the underlying mathematical operations tremendously through massive parallelism. Our contribution is performance characterization of multiple CNN-based models for object recognition and detection with several different hardware platforms and software frameworks, using both local (on-device) and remote (network-side server) computation. The measurements are conducted using real workloads and real processing platforms. On the platform side, we concentrate especially on TensorFlow and TensorRT. Our measurements include embedded processors found on mobile devices and high-performance processors that can be used on the network side of mobile systems. We show that there exists significant latency-throughput trade-offs but the behavior is very complex. We demonstrate and discuss several factors that affect the performance and yield this complex behavior.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Latency and throughput characterization of convolutional neural networks for mobile computer vision

Hanhirova

Kämäräinen

Seppälä

et al. 2018

Proceedings of the 9th ACM Multimedia Systems Conference

View full text Add to dashboard Cite

show abstract

“…Besides, a GPU-accelerated library called CNNdroid has been introduced. This library can execute trained CNN on Android-based mobile devices (Oskouei et al, 2015).…”

Section: Deep Neural Networkmentioning

confidence: 99%

Multipatch-GLCM for Texture Feature Extraction on Classification of the Colon Histopathology Images using Deep Neural Network with GPU Acceleration

Haryanto¹,

Pratama²,

Suhartanto³

et al. 2020

Journal of Computer Science

View full text Add to dashboard Cite

Cancer is one of the leading causes of death in the world. It is the main reason why research in this field becomes challenging. Not only for the pathologist but also from the view of a computer scientist. Hematoxylin and Eosin (H&E) images are the most common modalities used by the pathologist for cancer detection. The status of cancer with histopathology images can be classified based on the shape, morphology, intensity, and texture of the image. The use of full high-resolution histopathology images will take a longer time for the extraction of all information due to the huge amount of data. This study proposed advance texture extraction by multi-patch images pixel method with sliding windows that minimize loss of information in each pixel patch. We use texture feature Gray Level Co-Occurrence Matrix (GLCM) with a meanshift filter as the data pre-processing of the images. The mean-shift filter is a low-pass filter technique that considers the surrounding pixels of the images. The proposed GLCM method is then trained using Deep Neural Networks (DNN) and compared to other classification techniques for benchmarking. For training, we use two hardware: NVIDIA GPU GTX-980 and TESLA K40c. According to the study, Deep Neural Network outperforms other classifiers with the highest accuracy and deviation standard 96.72±0.48 for four cross-validations. The additional information is that training using Theano framework is faster than Tensorflow for both in GTX-980 and Tesla K40c.

show abstract

“…These GPUs have parallel processing capabilities which can be exploited to accelerate CNN computations on mobile devices. Moreover, an open source, GPU accelerated, library has recently become available on github [75]. Apart from this there is also a neural compute stick (Movidius) available on the market which shows promising results for the use of some CNN on low power devices [76].…”

Section: Existing Sbds Useful For Space Missionsmentioning

confidence: 99%

“The Smartphone’s Guide to the Galaxy”: In Situ Analysis in Space

2018

View full text Add to dashboard Cite

A human mission to Mars can be viewed as the apex of human technological achievement. However, to make this dream a reality several obstacles need to be overcome. One is devising practical ways to safeguard the crew health during the mission through the development of easy operable and compact sensors. Lately, several smartphone-based sensing devices (SBDs) with the purpose to enable the immediate sensitive detection of chemicals, proteins or pathogens in remote settings have emerged. In this critical review, the potential to piggyback these systems for in situ analysis in space has been investigated on application of a systematic keyword search whereby the most relevant articles were examined comprehensively and existing SBDs were divided into 4 relevant groups for the monitoring of crew health during space missions. Recently developed recognition elements (REs), which could offer the enhanced ability to tolerate those harsh conditions in space, have been reviewed with recommendations offered. In addition, the potential use of cell free synthetic biology to obtain long-term shelf-stable reagents was reviewed. Finally, a synopsis of the possibilities of combining novel SBD, RE and nanomaterials to create a compact sensor-platform ensuring adequate crew health monitoring has been provided.

show abstract

CNNdroid

Cited by 72 publications

References 8 publications

Latency and throughput characterization of convolutional neural networks for mobile computer vision

Latency and throughput characterization of convolutional neural networks for mobile computer vision

Multipatch-GLCM for Texture Feature Extraction on Classification of the Colon Histopathology Images using Deep Neural Network with GPU Acceleration

“The Smartphone’s Guide to the Galaxy”: In Situ Analysis in Space

Contact Info

Product

Resources

About