AI Powered Drones solving Your Real World Problems
Use of AI Drones in Solving Your Real World Problems
Drones & AI are like a match designed by the cosmic plan. More than ever before drones can play an extended role in helping us solve real world problems. With the advent of advanced programmable drones, like DJI Matrice Series, their usage in a variety of industries is a no surprise. It has become an essential tool for everyone from firefighters to farmers and livestock managers.
AI Drones & Folio3
No surprise that Folio3’s Machine Learning practise has been focusing on the usage of programmable AI drones, wherever required or possible. Drones have been only able to display what their cameras and sensors capture. With the advent of programmable AI drones they not only capture via the mounted camera but they can be programmed to act intelligently by observing their surroundings. This advanced computer vision capability enables them to map areas, track moving and static objects and provide a real time feedback.
Terrain Mapping System For Civil Engineers
Terrain mapping is required in a variety of industries for example in Agriculture, Civil engineering etc. One of our solutions involves terrain mapping for a company which is engaged in civil construction. They require to reconstruct a 3D model by using a CAD system. The problem description was to have a process in place which can take recorded or live feed from a 3D camera, and LIDAR, and then transform it into an adequate set of data in a canonical format such that any other CAD system can use it to reconstruct the world in a 3D model.
High Level Steps of the Solution Were:
A drone fitted with 3D camera and LIDAR detector will make the flight on the area of interest. The AI drone is either pre programmed on a pre-coordinated path or by using manual navigation
- The drone will follow the road/track with the help of external/pre-planned navigation.
- The drone will do the recording from its both sensors, and either broadcast the live stream, or record it in the on-board storage.
- The 3D camera data will be a video feed, whereas LIDAR data will be point cloud data.
- Both types of data will be geo tagged with proper timing information.
- The required system will take input from the drone.
- The system using the geo coordinates, and timing information to create mapping data based on point cloud, and textures from the 3D camera feed.
- This will provide altitudes, bumps, and objects on the surfaces.
- The texture will provide information on the object, and surface boundaries.
- The system, using the computer vision (CV) techniques, will identify boundaries of the roads.
- The system will record the prepared data in a format that can be used by specific third party CAD systems.
Our Featured Work
ATM Cash Forecasting
The system helps banks in avoiding both out of cash and over stock situations. It is a per-ATM cash outflow predictor, proposing LSTM-based deep learning models on time-series data.
Breast Cancer HER2 Subtype Identification
Our system provides an automated pipeline for cell segmentation and spot counting from a Computer Vision-based diagnostic-aid for the Fluorescent In-Situ Hybridization test.
Road Traffic Analysis
A Vehicle Counting System Based on Vehicle Type and Classification which is Using Deep Learning Method to differentiate different types of vehicles along with the total count.
Our system provides automated analysis of Gel Electrophoresis Images Predicting Thalassemia and tests for mutant gene expression involving fine-granularity medical image analysis.
Converse Smartly ®
Converse Smartly allows you to make your conversations smart, intelligent and productive with use of machine learning, artificial intelligence and Natural Language Processing.
Automated Authentication for Drive Thrus
Multi-Modal Automated Authentication for Drive-thrus via fine-grained car’s make/model classification and person identification using deep learning.
Facial Recognition System
A highly accurate facial recognition system which provides real time results, it is based on Histogram of Oriented Gradients (HOG) and Convolutional Neural Network (CNN).
Customer Churn Prediction
A solution based on Predictive modelling provides a data driven insight for businesses to identify potentially unhappy customers with business process validation.
Completion Time Estimation
Machine Learning powered solution which helps our client for an accurate and effective estimation of their digital marketing campaigns. It also simplifies the automation process for scheduling.