Deep Green Diagnostics: Urban Green Space Analysis Using Deep Learning and Drone Images

A fusion approach using GIS, green area detection, weather API and GPT for satellite image based fertile land discovery and crop suitability

Proper utilization of agricultural land is a big challenge as they often laid over as waste lands. Farming is a significant occupation in any country and improving it further by promoting more farming opportunities will take the country towards making a huge leap forward. The issue in achieving this would be the lack of knowledge of cultivable land for food crops. The objective of this work is to utilize modern computer vision technology to identify and map cultivable land for agricultural needs. With increasing population and demand for food, improving the farming sector is crucial. However, the challenge lies in the lack of suitable land for food crops cultivation. To tackle this issue, we propose to use sophisticated image processing techniques on satellite images of the land to determine the regions that are capable of growing food crops. The solution architecture includes enhancement of satellite imagery using sophisticated pan sharpening techniques, notably the Brovey transformation, aiming to transform dull satellite images into sharper versions, thereby improving the overall quality and interpretability of the visual data. Making use of the weather data on the location observed and taking into factors like the soil moisture, weather, humidity, wind, sunlight times and so on, this data is fed into a generative pre-trained transformer model which makes use of it and gives a set of crops that are suitable to be grown on this piece of land under the said conditions. The results obtained by the proposed fusion approach is compared with the dataset provided by the government for different states in India and the performance was measured. We achieved an accuracy of 80% considering the crop suggested by our model and the predominant crop of the region. Also, the classification report detailing the performance of the proposed model is presented.

Enhanced Automatic Identification of Urban Community Green Space Based on Semantic Segmentation

At the neighborhood scale, recognizing urban community green space (UCGS) is important for residential living condition assessment and urban planning. However, current studies have embodied two key issues. Firstly, existing studies have focused on large geographic scales, mixing urban and rural areas, neglecting the accuracy of green space contours at fine geographic scales. Secondly, the green spaces covered by shadows often suffer misclassification. To address these issues, we created a neighborhood-scale urban community green space (UCGS) dataset and proposed a segmentation decoder for HRNet backbone with two auxiliary decoders. Our proposed model adds two additional branches to the low-resolution representations to improve their discriminative ability, thus enhancing the overall performance when the high- and low-resolution representations are fused. To evaluate the performance of the model, we tested it on a dataset that includes satellite images of Shanghai, China. The model outperformed the other nine models in UCGS extraction, with a precision of 83.01, recall of 85.69, IoU of 72.91, F1-score of 84.33, and OA of 89.31. Our model also improved the integrity of the identification of shaded green spaces over HRNetV2. The proposed method could offer a useful tool for efficient UCGS detection and mapping in urban planning.

Mapping Urban Green Spaces at the Metropolitan Level Using Very High Resolution Satellite Imagery and Deep Learning Techniques for Semantic Segmentation

Urban green spaces (UGSs) provide essential environmental services for the well-being of ecosystems and society. Due to the constant environmental, social, and economic transformations of cities, UGSs pose new challenges for management, particularly in fast-growing metropolitan areas. With technological advancement and the evolution of deep learning, it is possible to optimize the acquisition of UGS inventories through the detection of geometric patterns present in satellite imagery. This research evaluates two deep learning model techniques for semantic segmentation of UGS polygons with the use of different convolutional neural network encoders on the U-Net architecture and very high resolution (VHR) imagery to obtain updated information on UGS polygons at the metropolitan area level. The best model yielded a Dice coefficient of 0.57, IoU of 0.75, recall of 0.80, and kappa coefficient of 0.94 with an overall accuracy of 0.97, which reflects a reliable performance of the network in detecting patterns that make up the varied geometry of UGSs. A complete database of UGS polygons was quantified and categorized by types with location and delimited by municipality, allowing for the standardization of the information at the metropolitan level, which will be useful for comparative analysis with a homogenized and updated database. This is of particular interest to urban planners and UGS decision-makers.

Modelling Accessibility to Urban Green Areas Using Open Earth Observations Data: A Novel Approach to Support the Urban SDG in Four European Cities

Cities are recognised as key enablers for the world’s sustainable future. Urban sprawl and inefficient use of land are important issues significantly impacting the provision and use of open green spaces. The United Nations Sustainable Development Goal (SDG) indicator 11.7.1 aims at globally monitoring the amount of land that is dedicated by cities for public space. In Europe, the indicator “Share of urban population without green urban areas in their neighbourhood” is supposed to correspond to the SDG11.7.1 but is currently on-hold due to methodological issues and lack of data. Moreover, to efficiently assess public space conditions, timely and spatially disaggregated information is essential but not yet widely adopted by urban practitioners. Hereafter, we use a combination of satellite and crowdsourced Earth Observations (EO) to model physical accessibility to urban green spaces in four European cities. Findings suggest that it is technically feasible to derive information on the share of urban population without green urban areas in their neighbourhood. Results demonstrate that the proposed methodology represents a consistent, valid, reliable, low-cost, timely and continuous source of information for sustainable urban development. Open and free EO data can be a good complement to enhance official and traditional statistics on urban areas facilitating EU reporting against the SDG indicator for better comparison between EU countries.