Raster-Zoo is part of the PredictiveWorks. project family and aims to seamlessly integrate imagery data from drones and satellites with deep learning. Use cases are image classification (including location), segmentation and object detection.
For the described goal, Raster-Zoo integrates two outstanding open source projects: RasterFrames from Astraae and, Analytics-Zoo from Intel. This results in configurable, code-free deep learning pipelines for training and inference phases.
PredictiveWorks. is a declarative AI factory to turn AI solutions from artisanal hand-crafted AI products into industrial assets, that can be produced on demand and just in time.
From a functional perspective, PredictiveWorks. is made to support the entire AI triad, from data to models and solutions with a single platform and technology.
From a data perspective, PredictiveWorks. has a strong focus on the combination of sensor, imagery and security data (i.e. endpoint and traffic data) to support multi-step AI applications, from smart & secure agriculture, city, industry to and energy.
This open source library is an enabler for global-scale deep and machine learning. It allows data scientists, analysts and software developers to process and analyze geospatial-temporal raster data with the same flexibility and ease as any other data type in Apache Spark DataFrames. RasterFrames adds raster data support to the Apache Spark ecosystem.
This complements Internet of Things sensor readings and Cyber Defense endpoint & traffic data to reach full situational awareness for secure (industrial) IoT environments.
Analytics Zoo seamless scales TensorFlow, Keras and PyTorch to distributed big data based on Apache Spark, Flink and Ray. It lets you easily apply existing AI models (from TensorFlow to BigDL to OpenVINO etc.), build your own AI models from JSON configurations without writing a single line of code and automate the deep & machine learning process (like feature engineering, hyper parameter tuning, model selection, distributed inference and more).