CSS Corp is enabling leading mapping companies to accelerate POI extraction from geotagged images through AI-ML led automation
While navigating using a mobile device, we often select a source and a destination that are well-known and frequently visited places, such as hotels, apartment complexes, tourist attractions, corporate offices, etc. In mapping parlance, these places are called points of interest (POI). Most location-based applications and services need accurate POI data to serve their users effectively. Among several ways to capture POI data, extracting it from geotagged images is one of the most popular. Geotagged images contain geographical metadata like latitude, longitude, and place names, etc.
However, mapping companies often find it challenging to get detailed POI data generated from geotagged images accurately. Fierce competition in this field has created a demand for high quality and freshness of POI data. It necessitates using efficient processes that bring reality to the maps in real-time or as soon as possible.
A leading mapping and location data platform provider was under immense pressure to scale their services and capture the market share rapidly. A critical component of their services was seamless POI extraction from field-collected geotagged images while maintaining quality and accuracy benchmarks. Done manually, this process can be tedious and time-consuming. CSS Corp was able to support them through rapid deployment of trained resources at scale, empowered with an assisted automation approach, that accelerated the time-to-market for their services. It leveraged its proprietary Geo.Intelli system which uses artificial intelligence for automated extraction of POI data from geotagged images, resulting in faster and efficient processing.
Geo.Intelli is a smart GIS system that automates the geotag extraction for POI location from images and leverages NLP to check the completeness of POI or address name. Its AI-ML based APIs automatically extract the relevant data from images, perform a quality check on images, and reject images that are blurred, non-geocoded, or in invalid format. To ensure high accuracy, the system automatically cross-validates the extracted data with reference source data like area, city, latitude, longitude, and ZIP Code. It also allows for multiple POI addition from a single image.