
Resources
Title:
Spread: A Large-Scale, High-Fidelity Synthetic Dataset for Multiple Forest Vision Tasks
Authors:
Zhengpeng (Frank) Feng, Yihang She, EEG Group, PhD students and Prof Srinivasan Keshav in the Department of Computer Science and Technology at the University of Cambridge.
Description:
We present the Synthetic Photo-realistic Arboreal Dataset (SPREAD), a state-of-the-art synthetic dataset specifically designed for forest-related machine-learning tasks. Developed using Unreal Engine 5, SPREAD goes beyond existing synthetic forest datasets in terms of realism, diversity, and comprehensiveness. It includes RGB, depth images, point clouds, semantic and instance segmentation labels, along with key parameters such as tree ID, location, diameter at breast height (DBH), height, and canopy diameter. In exemplary experiments, we found that SPREAD significantly reduces the need to use real-world datasets for trunk segmentation tasks and enhances model segmentation performance. Specifically, by pretraining on SPREAD, MobileNetV3 and DeepLabV3 models require only 25% of a fine-tuning real-world dataset to match or even surpass the performance of ImageNet-pretrained models fine-tuned on the entire real-world dataset. Furthermore, our hybrid training experiments demonstrate that combining SPREAD and real data at a 1:1 or 2:1 ratio greatly improves task performance. For the canopy instance segmentation task, SPREAD pretraining still provides varying degrees of performance improvement for the models. All datasets, data collection frameworks, and codes are available at https://github.com/FrankFeng-23/SPREAD.
Title:
SPAGHETTI: A Synthetic Data Generator for post-Covid Electric Vehicle Usage
Authors:
Anaïs-Marie Celestine Berkes, EEG Group, PhD student and Gates Scholar, and Prof Srinivasan Keshav in the Department of Computer Science and Technology at the University of Cambridge.
Description:
SPAGHETTI (Synthetic Patterns & Activity Generator for Home-Energy & Tomorrow’s Transportation Investigation) is a tool that can be used for the synthetic generation of realistic EV drive cycles. It takes as input EV user commuting and non-commuting patterns, allowing for personalised modeling of EV usage. It is based on a thorough literature survey on post-Covid work-from-home (WFH) patterns and can also be used to model the EV usage patterns of the three most common types of post-Covid workers. SPAGHETTI can be used by the scientific community to conduct further research on the large-scale adoption of EVs and their integration into domestic microgrids.
Please click on the link for more details.
Title:
SOPEVS: Sizing and Operation of PV-EV-Integrated Modern Homes
Authors:
Anaïs-Marie Celestine Berkes, EEG Group, PhD student and Gates Scholar, and Prof Srinivasan Keshav in the Department of Computer Science and Technology at the University of Cambridge.
Description:
We address a problem that arises at the confluence of three recent trends: the popularity of storage-coupled photovoltaic (PV) systems amongst homeowners, the rapid proliferation of electric vehicles (EVs) with potential for bidirectional energy storage within PV-enabled single-family homes, and third, the surge in remote working accelerated by the Covid-19 pandemic. In this context, we explore the joint optimal sizing and operation of domestic homes while accounting for different degrees of remote working and the impact of home energy management system (HEMS) operation preferences. This task is complex due to the coupling between sizing and operation and the stochastic and non-stationary nature of solar generation, load, and EV drive cycles. We introduce SOPEVS (Sizing & Operation of PV and EV integrated Single-family homes), a novel framework formulated to tackle this multifaceted challenge. We use SOPEVS to investigate how commuting habits and choices in HEMS operation affect the sizing of domestic PV energy systems. Our findings reveal that homeowners who predominantly work from home and possess bidirectional EVs can potentially eliminate the need for separate home storage systems, thereby substantially reducing overall system costs. We also find that configuring a HEMS to maximise charging through solar energy can achieve savings of up to 80% on total system expenditure (excluding the cost of EV), depending on the desired level of grid independence and the preferred State of Charge (SOC) of EV at the time of departure.
Please click on the link for more details and the tool links.
Title:
Heatalyzer: A Tool for Evaluating Indoor Comfort in Buildings during Extreme Heat Events
Authors:
Livia Capol, Master’s student in Computer Science at ETH Zurich and Visiting Student, and Prof Srinivasan Keshav at the Department of Computer Science and Technology at the University of Cambridge
Description:
The increase in heatwaves due to climate change poses significant challenges to both indoor thermal comfort and occupant well-being. Unfortunately, existing work does not quantify the impact of extreme heat events as a function of building type, occupant age, and heatwave intensity and duration. We therefore present Heatalyzer, a novel Building Energy Modeling (BEM) tool to analyze indoor thermal comfort, liveability, and survivability for a range of buildings under both past and future weather scenarios. It lets users compare these outcomes across building types and weather scenarios, integrates algorithms for creating extreme weather data, and has a user-friendly interface. Additionally, it outputs several commonly used thermal comfort metrics, which are crucial for evaluating indoor conditions during heatwaves. These functionalities enable a wide audience, including building managers and policymakers, to assess the impact of extreme heat events on building occupants.
Please click on the link for more details.
Image under Pexels License.