Computer vision based long range object detection, classification & tracking.
AI & DATA SCIENCE
AI & data science
The successful integration of data, algorithms and intelligence to products brings new technological challenges. Starting with challenges to get the right data, proving feasibility with minimal investment, finding the right expertise development and successful deployment into production.
Key challenges for AI & data science
Depth of AI technology
Developing the right algorithm for the job requires a versatile team mastering all subdomains of AI, algorithms and data science.
Integrated system design
Algorithms need to be matched precisely to the use case, compute platform, sensors and data sources. Design, select and integrate these on a system level with a multidisciplinary approach.
Data gathering & management
Determining data requirements for big and small data projects. Provide end-to-end services for gathering, data labeling, and data management.
Deployment & production
Developing algorithms for a specific target device: embedded electronics, mobile devices or cloud applications, so the algorithms are tailored for smooth and frugal operations on any device.
Integrating specific technology
Bringing artificial intelligence to your product isn’t about the technology itself, it’s about getting maximum value from your data, whatever the amount, type or context. Verhaert helps you to gather data, select the right sensors, develop the right algorithm and optimize it for any target platform.
Machine Learning
- Custom deep learning models
- (Un)supervised & self-supervised ML
- Surrogate model development
- Bayesian networks & graphs
- AutoML, active learning
- ML & physics: hybrid models
- MLops in production
- ML optimized for edge devices
- ML optimized for cloud
Optimization
- Resource planning & optimization
- Graph optimization
- Bayesian optimization
- Evolutionary algorithms
- Swarm optimization
- Stochastic optimization
- Constraint optimization
- Optimal routing
Robotics AI
- Probabilistic robotics: planning & controlling robots under uncertainty
- Automated path planning & navigation
- Motion planning & control
- Perception & situational awareness
- SLAM & VSLAM
- Sensor fusion
- Robot programming & simulation (ROS)
Structured & tabular data algorithms
- Time series forecasting, prediction & classification
- Anomaly detection
- Automated pattern finding
- Recommendation algorithms
- Preference mining
- Clustering
- Business intelligence algorithms
Computer vision
- Image & video processing
- Image classification, segmentation, object detection & tracking
- Video analysis
- Image generation
- Lidar & radar based computer vision
- Computer vision: stitching, registration, depth estimation
Data science & statistics
- Automatic labeling techniques
- Statistical analysis
- Data visualization: power BI, grafana, …
- Databases: (No)SQL, …
- Cloud data & processing pipelines: AWS, Google, IBM, AZURE
- Data cleaning: missing data estimation
Natural Language Processing (NLP)
- Text based: classification, topic extraction, …
- Speech recognition & voice control
- Knowledge extraction & information retrieval
- Text generation
- Chatbots (voice & text)
- Information retrieval
Advanced control
- ML based system identification & surrogate models
- Reinforcement learning based closed loop control
- Advanced control: model predictive control, …
- Advanced closed loop control
Technology building blocks
Machine learning model repository
IRIS – Data gathering & visualization engine
Robotics planning & advanced control engine
AI optimization library
Get in touch
Verhaert’s mature technologies are accessible and ready to be implemented to make your products more autonomous, better performing and easier to use. Want to talk about the implementation for your products?
