We present a class of efficient models called MobileNets for mobile and embedded vision applications. MobileNets are based on a streamlined architecture that uses depthwise separable convolutions to build light weight deep neural networks. We introduce two simple global hyperparameters that efficiently trade off between latency and accuracy. These hyper-parameters allow the model builder to choose the right sized model for their application based on the constraints of the problem. We present extensive experiments on resource and accuracy tradeoffs and show strong performance compared to other popular models on ImageNet classification. We then demonstrate the effectiveness of MobileNets across a wide range of applications and use cases including object detection, finegrain classification, face attributes and large scale geo-localization.

We present the next generation of MobileNets based on a combination of complementary search techniques as well as a novel architecture design. MobileNetV3 is tuned to mobile phone CPUs through a combination of hardwareaware network architecture search (NAS) complemented by the NetAdapt algorithm and then subsequently improved through novel architecture advances. This paper starts the exploration of how automated search algorithms and network design can work together to harness complementary approaches improving the overall state of the art. Through this process we create two new MobileNet models for release: MobileNetV3-Large and MobileNetV3-Small which are targeted for high and low resource use cases. These models are then adapted and applied to the tasks of object detection and semantic segmentation. For the task of semantic segmentation (or any dense pixel prediction), we propose a new efficient segmentation decoder Lite Reduced Atrous Spatial Pyramid Pooling (LR-ASPP). We achieve new state of the art results for mobile classification, detection and segmentation. MobileNetV3-Large is 3.2% more accurate on ImageNet classification while reducing latency by 15% compared to MobileNetV2. MobileNetV3-Small is 4.6% more accurate while reducing latency by 5% compared to MobileNetV2. MobileNetV3-Large detection is 25% faster at roughly the same accuracy as MobileNetV2 on COCO detection. MobileNetV3-Large LR-ASPP is 30% faster than MobileNetV2 R-ASPP at similar accuracy for Cityscapes segmentation.

Coronavirus disease 2019 (COVID-19) most commonly presents with respiratory symptoms, including cough, shortness of breath, and sore throat. However, digestive symptoms also occur in patients with COVID-19 and are often described in outpatients with less severe disease. In this study, we sought to describe the clinical characteristics of COVID-19 patients with digestive symptoms and mild disease severity. METHODS:We identified COVID-19 patients with mild disease and one or more digestive symptoms (diarrhea, nausea, and vomiting), with or without respiratory symptoms, and compared them with a group presenting solely with respiratory symptoms. We followed up patients clinically until they tested negative for COVID-19 on at least 2 sequential respiratory tract specimens collected ‡24 hours apart. We then compared the clinical features between those with digestive symptoms and those with respiratory symptoms. RESULTS:There were 206 patients with low severity COVID-19, including 48 presenting with a digestive symptom alone, 69 with both digestive and respiratory symptoms, and 89 with respiratory symptoms alone. Between the 2 groups with digestive symptoms, 67 presented with diarrhea, of whom 19.4% experienced diarrhea as the first symptom in their illness course. The diarrhea lasted from 1 to 14 days, with an average duration of 5.4 6 3.1 days and a frequency of 4.3 6 2.2 bowel movements per day. Concurrent fever was found in 62.4% of patients with a digestive symptom. Patients with digestive symptoms presented for care later than those with respiratory symptoms (16.0 6 7.7 vs 11.6 6 5.1 days, P < 0.001). Nevertheless, patients with digestive symptoms had a longer duration between symptom onset and viral clearance (P < 0.001) and were more likely to be fecal virus positive (73.3% vs 14.3%, P 5 0.033) than those with respiratory symptoms. DISCUSSION:We describe a unique subgroup of COVID-19 patients with mild disease severity marked by the presence of digestive symptoms. These patients are more likely to test positive for viral RNA in stool, to have a longer delay before viral clearance, and to experience delayed diagnosis compared with patients with only respiratory symptoms.

Purpose The purpose of this paper is to examine the effects of perceived benefits, i.e. utilitarian value, hedonic value and social value, as well as perceived risk, on purchase intention in social commerce context. Design/methodology/approach To cast light on the factors motivating users’ intention to purchase in the context of social commerce, data of 277 users of social commerce in China were collected via an online survey. Findings Results show that satisfaction significantly and positively affects users’ purchase intention in social commerce context. In addition, utilitarian, hedonic and social values have significant and positive impacts on satisfaction and purchase intention; and utilitarian value is found to be the most salient factor influencing purchase intention, while hedonic value has the greatest effect on satisfaction. Moreover, perceived risk significantly and negatively affects satisfaction. Originality/value Extant research on social commerce has mainly focused on investigating how the general perceived value affects user behavior, but has less considered different dimensions of perceived value. Moreover, prior studies have explored the roles of utilitarian and hedonic values on user behavior; however, there is a lack of research on the effect of social value. The current study attempts to fill these research gaps.

No abstract