Recent research has demonstrated the potential of deploying deep neural networks (DNNs) on resource-constrained mobile platforms by trimming down network complexity using different compression techniques. The current practice only investigate stand-alone schemes even though each technique may be well suited for certain types DNN layers. Also, these techniques are optimized merely inference accuracy DNNs, without explicitly considering other application-driven system performance (e.g. latency...

Non-speech sound-awareness is important to improve the quality of life for deaf and hard-of-hearing (DHH) people. DHH people, especially young, are not always satisfied with their hearing aids. According interviews 60 young students, a ubiquitous tool emergency social events that works in diverse environments desired. In this paper, we design UbiEar, smartphone-based acoustic event sensing notification system. Core techniques UbiEar light-weight deep convolution neural network enable...

Despite the rapid development of mobile and embedded hardware, directly executing computation-expensive storage-intensive deep learning algorithms on these devices' local side remains constrained for sensory data analysis. In this paper, we first summarize layer compression techniques state-of-the-art model from three categories: weight factorization pruning, convolution decomposition, special architecture designing. For each category techniques, quantify their storage computation tunable by...

Recent breakthroughs in deep neural networks (DNNs) have fueled a tremendously growing demand for bringing DNN-powered intelligence into mobile platforms. While the potential of deploying DNNs on resource-constrained platforms has been demonstrated by DNN compression techniques, current practice suffers from two limitations: 1) merely stand-alone schemes are investigated even though each technique only suit certain types layers; and 2) mostly techniques optimized DNNs' inference accuracy,...

The following topics are dealt with: geophysical image processing; remote sensing; classification; time series; learning (artificial intelligence); feature extraction; radar imaging; synthetic aperture radar; resolution; convolutional neural nets.

Acoustic alarms have been credited with saving thousands of lives from fires, gas leakage and electric each year. By broadcasting sound different tones, loudness timbres, acoustic keep people aware surroundings, inform them serendipitous events, notify critical information. However, maintaining the safety awareness through alarm is difficult for who are deaf or less sensitive to signals. They too often among last access important information even when they in dangers, especially stay alone....

The atmospheric environment is facing increasing threats from industrial pollutions. This paper presents an air pollution source estimation algorithm using mobile sensor networks. We propose a continuous point model of under windy conditions. Then we use quadrocopters which equipped with sensors that can detect pollutants to collect concentration information. Based on the collected information, take advantage maximum likelihood method estimate diffusion parameters. To improve accuracy...

The information gradient-based routing and navigation protocols have been proved to be effective when collecting data from the distributed wireless sensor networks, because collector can achieve global objective through local greedy decision based on gradient. An efficient method establish this gradient is solve a discrete approximation harmonic function, which called potential field (IPF). However, energy-efficient quick convergence methods construct IPF should fully investigated trade off...

Lack of physical activity is becoming a killer our healthy life. As solution for this negative impact, we propose SmartCare to help users set up habit. can monitor user's activities over long time, and then provide quality assessment suggestion. consists three parts, recognition, energy saving, health feedback. Activity recognition recognize nine kinds daily activities. A hybrid classifier that uses less power memory with satisfactory accuracy was designed implemented by utilizing the...

The information gradient-based routing protocols have been proved to be economical and effective by adopting the principle of achieving global objective through local decision, but lightweight methods construct gradient should fully investigated, especially in a large-scale network with high dynamics. In this paper, we focus on construction balancing convergence conditions energy consumption. Therefore, two algorithms, Hierarchical Skeleton-based Construction Algorithm (HSCA) Estimate value...

Abstract. Olivine is a mineral indicative of water activity in the mantle, and its presence can elucidate petrological evolution history magmatic igneous rocks. In this work, we conducted preliminary investigation into characteristics patterns spatial relationships between representative Martian (i.e., olivine) geomorphic features, aiming to provide geographical insights potential habitable environment Mars. Specifically, an analytical framework constructed explore connection hydrated...

Recent breakthroughs in Deep Neural Networks (DNNs) have fueled a tremendously growing demand for bringing DNN-powered intelligence into mobile platforms. While the potential of deploying DNNs on resource-constrained platforms has been demonstrated by DNN compression techniques, current practice suffers from two limitations: 1) merely stand-alone schemes are investigated even though each technique only suit certain types layers; and 2) mostly techniques optimized DNNs' inference accuracy,...

Animal resources are significant to human survival and development the ecosystem balance. Automated multi-animal object detection is critical in animal research conservation monitoring. The objective design a model that mitigates challenges posed by large number of parameters computations existing methods. We developed backbone network with enhanced representative capabilities pursue this goal. This combines foundational structure Transformer Large Selective Kernel (LSK) module, known for...

Visual sensors on autonomous vehicles are vulnerable to adverse weather, which seriously reduces the performance of semantic segmentation task and threats people's safety. Therefore, models need carry out extensive training a large amount weather data is difficult expensive acquire improve their robustness. To solve this problem, researchers have proposed domain generalization methods that do not target (such as weather) adapt during training. However, most them focus synthetic-to-real...

Context model is widely used in image coding system to improve the compression performance. In order achieve complex high-level statistical correlation of source effectively, context can be get conditional probability current coded symbol. However, it proved that difficult true distribution signal, as a result effect reduced, which so-called cost problem. quantization an efficient method deal with this As quantification similar general vector problem, achieved by clustering algorithm under...

Macaque monkey is a rare substitute which plays an important role for human beings in relation to psychological and spiritual science research. It essential these studies accurately estimate the pose information of macaque monkeys. Many large-scale models have achieved state-of-the-art results estimation. However, it difficult deploy when computing resources are limited. Combining structure high-resolution network design principle light-weight network, we propose attention-refined estimation...

Deep neural networks (DNNs) play an important role in a variety of intelligent applications (e.g. image classification and target recognition), yet at the cost heavy computation burden, that makes DNNs difficult to deploy on resource-constrained IoT devices. To solve this problem, there are two categories model adjustment methods: compression segmentation. However, mainly reduces resource consumption accuracy while segmentation according communication latency. In paper, we propose Joint...