This paper is about few-shot segmentation of foreground objects in images. We train a CNN on small subsets of training images, each mimicking the few-shot setting. In each subset, one image serves as the query and the other(s) as support image(s) with ground-truth segmentation. The CNN first extracts feature maps from the query and support images. Then, a class feature vector is computed as an average of the support's feature maps over the known foreground. Finally, the target object is segmented in the query image by using a cosine similarity between the class feature vector and the query's feature map. We make two contributions by: (1) Improving discriminativeness of features so their activations are high on the foreground and low elsewhere; and (2) Boosting inference with an ensemble of experts guided with the gradient of loss incurred when segmenting the support images in testing. Our evaluations on the PASCAL-5 i and COCO-20 i datasets demonstrate that we significantly outperform existing approaches.

The aim of this study was to disaggregate water flow data collected from high resolution smart water meters into different water end use categories. The data was obtained from a sample of 252 residential dwellings located within South East Queensland (SEQ), Australia.An integrated approach was used, combining high resolution water meters, remote data transfer loggers, household water appliance audits and a self-reported household water use diary. Disaggregating water flow traces into a registry of end use events (e.g. shower, clothes washer, etc.) is predominately a complex pattern matching problem, which requires a comparison between presented patterns and those contained with a large registry of categorised end use events. Water flow data collected directly from water meters includes both single (e.g. shower event occurring alone) and combined events (i.e. an event which comprises of several overlapped single events). To identify these former mentioned single events, a hybrid combination of the Hidden Markov Model (HMM) and the Dynamic Time Warping Algorithm (DTW) provided the most feasible and accurate approach available.Additional end use event physical context algorithms have been developed to aid accurate end use event categorisation. This paper firstly presents a thorough discussion on the single water end use event analysis process developed and its internal validation with a testing set. This is followed by the application of the developed approach on three independent households to examine its degree of accuracy in disaggregating two weeks of residential flow data into a repository of residential water end use events. Future stages of algorithm development and testing is discussed in the final section.

The rapid dissemination of residential water end-use (e.g. shower, clothes washer, etc.) consumption data to the customer via a web-enabled portal interface is becoming feasible through the advent of high resolution smart metering technologies. However, in order to achieve this paradigm shift in residential customer water use feedback, an automated approach for disaggregating complex water flow trace signatures into a registry of end-use event categories needs to be developed. This outcome is achieved by applying a hybrid combination of gradient vector filtering, Hidden Markov Model (HMM) and Dynamic Time Warping Algorithm (DTW) techniques on an existing residential water end-use database of 252 households located in Southeast Queensland, Australia having high resolution water meters (0.0139 L/pulse), remote data transfer loggers (5s logging) and completed household water appliance audits. The approach enables both single independent events (e.g. shower event) and combined events (i.e. several overlapping single events) to be disaggregated from flow data into a comprehensive end-use event registry. Complex blind source separation of concurrently occurring water end use events (e.g. shower and toilet flush occurring in same time period) is the primary focus of this present study. Validation of the developed model is achieved through an examination of 50 independent combined events.

a b s t r a c tOver half of the world's population will live in urban areas in the next decade, which will impose significant pressure on water security. The advanced management of water resources and their consumption is pivotal to maintaining a sustainable water future. To contribute to this goal, the aim of this study was to develop an autonomous and intelligent system for residential water end-use classification that could interface with customers and water business managers via a user-friendly web-based application. Water flow data collected directly from smart water metres connected to dwellings includes both single (e.g., a shower event occurring alone) and combined (i.e., an event that comprises several overlapping single events) water end use events. The authors recently developed an intelligent application called Autoflow which served as a prototype tool to solve the complex problem of autonomously categorising residential water consumption data into a registry of single and combined events. However, this first prototype application achieved overall recognition accuracy of 85%, which is not sufficient for a commercial application. To improve this accuracy level, a larger dataset consisting of over 82,000 events from over 500 homes in Melbourne and South-east Queensland, Australia, were employed to derive a new single event recognition method employing a hybrid combination of Hidden Markov Model (HMM), Artificial Neural Networks (ANN) and the Dynamic Time Warping (DTW) algorithm. The classified single event registry was then used as the foundations of a sophisticated hybrid ANN-HMM combined event disaggregation module, which was able to strip apart concurrently occurring end use events. The new hybrid model's recognition accuracy ranged from 85.9% to 96.1% for single events and 81.8-91.5% for combined event disaggregation, which was a 4.9% and 8.0% improvement, respectively, when compared to the first prototype model. The developed Autoflow tool has far-reaching implications for enhanced urban water demand planning and management, sustained customer behaviour change through more granular water conservation awareness, and better customer satisfaction with water utility providers.

Current practice for the design of an urban water system usually relies on various models that are often founded on a number of assumptions on how bulk water consumption is attributed to customer connections and outdated demand information that does not reflect present consumption trends; meaning infrastructure is often unnecessarily overdesigned. The recent advent of high resolution smart water meters and advanced data analytics allow for a new era of using the continuous 'big data' generated by these meter fleets to create an intelligent system for urban water management to overcome this problem. The aim of this research is to provide infrastructure planners with a detailed understanding of how granular data generated by an intelligent water management system (Autoflow©) can be utilised to obtain significant efficiencies throughout different stages of an urban water cycle, from supply, distribution, customer engagement, and even wastewater treatment.