Construction of a geospatial labeling tool for AI dataset generation

Project Overview: The client operated a web and mobile application designed for image labeling (PNG and JPG formats) to create datasets for AI models. They enlisted me and two other developers to enhance the web application's capabilities by adding functionalities for annotating over maps. This new feature included the integration of satellite images and tile maps to serve as base maps for user annotations. Given my previous experience in geospatial data development, I led this expansion. Key Contributions: Project Creation: I spearheaded the development of a feature allowing the creation of projects using a variety of geospatial data and formats. This enabled users to leverage diverse datasets for more comprehensive analysis and mapping. Base Map Generation: I facilitated the integration with satellite image providers like Planet and Sentinel. This integration allowed us to procure satellite imagery that was then utilized as base maps for annotations. Additionally, I integrated proprietary tile maps to enrich the mapping features available to users. Annotation Workflow: I was instrumental in planning and implementing the annotation process for both the mobile and web platforms. My role was crucial in developing the entire backend for the mobile application, ensuring a seamless user experience across devices. Data Cleaning and Result Generation: Depending on the project requirements, I conducted extensive data cleaning processes and generated results in GIS formats. This was vital in ensuring the accuracy and usability of the data produced by our application. This project not only enhanced the functionality of our client's application but also significantly contributed to the efficiency of their data annotation processes, particularly in the geospatial context.