-
Efficient and Accurate Quantized Image Super-Resolution on Mobile NPUs, Mobile AI & AIM 2022 challenge: Report
Authors:
Andrey Ignatov,
Radu Timofte,
Maurizio Denna,
Abdel Younes,
Ganzorig Gankhuyag,
**gang Huh,
Myeong Kyun Kim,
Kihwan Yoon,
Hyeon-Cheol Moon,
Seungho Lee,
Yoonsik Choe,
**woo Jeong,
Sungjei Kim,
Maciej Smyl,
Tomasz Latkowski,
Pawel Kubik,
Michal Sokolski,
Yujie Ma,
Jiahao Chao,
Zhou Zhou,
Hongfan Gao,
Zhengfeng Yang,
Zhenbing Zeng,
Zhengyang Zhuge,
Chenghua Li
, et al. (71 additional authors not shown)
Abstract:
Image super-resolution is a common task on mobile and IoT devices, where one often needs to upscale and enhance low-resolution images and video frames. While numerous solutions have been proposed for this problem in the past, they are usually not compatible with low-power mobile NPUs having many computational and memory constraints. In this Mobile AI challenge, we address this problem and propose…
▽ More
Image super-resolution is a common task on mobile and IoT devices, where one often needs to upscale and enhance low-resolution images and video frames. While numerous solutions have been proposed for this problem in the past, they are usually not compatible with low-power mobile NPUs having many computational and memory constraints. In this Mobile AI challenge, we address this problem and propose the participants to design an efficient quantized image super-resolution solution that can demonstrate a real-time performance on mobile NPUs. The participants were provided with the DIV2K dataset and trained INT8 models to do a high-quality 3X image upscaling. The runtime of all models was evaluated on the Synaptics VS680 Smart Home board with a dedicated edge NPU capable of accelerating quantized neural networks. All proposed solutions are fully compatible with the above NPU, demonstrating an up to 60 FPS rate when reconstructing Full HD resolution images. A detailed description of all models developed in the challenge is provided in this paper.
△ Less
Submitted 7 November, 2022;
originally announced November 2022.
-
Artificial Neural Networks for Sensor Data Classification on Small Embedded Systems
Authors:
Marcus Venzke,
Daniel Klisch,
Philipp Kubik,
Asad Ali,
Jesper Dell Missier,
Volker Turau
Abstract:
In this paper we investigate the usage of machine learning for interpreting measured sensor values in sensor modules. In particular we analyze the potential of artificial neural networks (ANNs) on low-cost micro-controllers with a few kilobytes of memory to semantically enrich data captured by sensors. The focus is on classifying temporal data series with a high level of reliability. Design and im…
▽ More
In this paper we investigate the usage of machine learning for interpreting measured sensor values in sensor modules. In particular we analyze the potential of artificial neural networks (ANNs) on low-cost micro-controllers with a few kilobytes of memory to semantically enrich data captured by sensors. The focus is on classifying temporal data series with a high level of reliability. Design and implementation of ANNs are analyzed considering Feed Forward Neural Networks (FFNNs) and Recurrent Neural Networks (RNNs). We validate the developed ANNs in a case study of optical hand gesture recognition on an 8-bit micro-controller. The best reliability was found for an FFNN with two layers and 1493 parameters requiring an execution time of 36 ms. We propose a workflow to develop ANNs for embedded devices.
△ Less
Submitted 15 December, 2020;
originally announced December 2020.
-
Deuteron frozen spin polarized target for nd experiements at the VdG accelerator of Charles University
Authors:
N. S. Borisov,
N. A. Bazhanov,
A. A. Belyaev,
J. Broz,
J. Cerny,
Z. Dolezal,
A. N. Fedorov,
G. M. Gurevich,
M. P. Ivanov,
P. Kodys,
P. Kubik,
E. S. Kuzmin,
A. B. Lazarev,
F. Lehar,
O. O. Lukhanin,
V. N. Matafonov,
A. B. Neganov,
I. L. Pisarev,
J. Svejda,
S. N. Shilov,
Yu. A. Usov,
I. Wilhelm
Abstract:
A frozen spin polarized deuteron target cooled by the 3He/4He dilution refrigerator is described. Fully deuterated 1,2-propanediol was used as a target material. Deuteron vector polarization about 40% was obtained for the target in the shape of a cylinder of 2 cm diameter and 6 cm length. The target is intended for a study of 3N interactions at the polarized neutron beam generated by the Van de…
▽ More
A frozen spin polarized deuteron target cooled by the 3He/4He dilution refrigerator is described. Fully deuterated 1,2-propanediol was used as a target material. Deuteron vector polarization about 40% was obtained for the target in the shape of a cylinder of 2 cm diameter and 6 cm length. The target is intended for a study of 3N interactions at the polarized neutron beam generated by the Van de Graaff accelerator at the Charles University in Prague.
△ Less
Submitted 8 December, 2007;
originally announced December 2007.
-
Measurement of spin-dependent total cross-section difference $Δσ_T$ in neutron-proton scattering at 16 MeV
Authors:
J. Broz,
J. Cerny,
Z. Dolezal,
G. M. Gurevich M. Jirasek,
P. Kubik,
A. A. Lukhanin,
J. Svejda,
I. Wilhelm,
N. S. Borisov,
Yu. M. Kazarinov B. A. Khachaturov,
E. S. Kuzmin,
V. N. Matafonov,
A. B. Neganov,
I. L. Pisarev,
Yu. A. Plis,
Yu. A. Usov,
M. Rotter,
B. Sedlak
Abstract:
A new measurement of $Δσ_T$ for polarized neutrons transmitted through a polarized proton target at 16.2 MeV has been made. A polarized neutron beam was obtained from the $^{3}\rm{H}(d,\vec n)^{4}\rm{He}$ reaction; proton polarization over 90\% was achieved in a frozen spin target of 20 cm$^3$ volume. The measurement yielded the value $Δσ_T=(-126\pm21\pm14)$ mb. The result of a simple phase shif…
▽ More
A new measurement of $Δσ_T$ for polarized neutrons transmitted through a polarized proton target at 16.2 MeV has been made. A polarized neutron beam was obtained from the $^{3}\rm{H}(d,\vec n)^{4}\rm{He}$ reaction; proton polarization over 90\% was achieved in a frozen spin target of 20 cm$^3$ volume. The measurement yielded the value $Δσ_T=(-126\pm21\pm14)$ mb. The result of a simple phase shift analysis for the $^3S_1-^3D_1$ mixing parameter $ε_1$ is presented and compared with the theoretical potential model predictions.
△ Less
Submitted 28 July, 1995;
originally announced July 1995.
