Skip to content

AI & DATA SOFTWARE PLATFORM

Hearth®

Turn motion into intelligence

Develop, integrate and deploy Event-Based Vision applications on embedded systems with Hearth, Prophesee’s next-generation software and AI platform.

Built on a modular microservices-centric architecture designed for scalability, Hearth is an AI & compute platform bringing Prophesee-optimized algorithms, AI and application components across sensor, edge and cloud environments to help you get the most from your Event-Based Vision application.

Inquire
Event-based views processed by Hearth: rail infrastructure, drone tracking and particle analysis

Trusted and sovereign by design

Industrial-quality software with cybersecurity and compliance addressed from sensor to decision, including CRA and EU AI Act requirements, sovereign deployment options, AI governance and auditability for critical decisions.

Robust multimodal perception

Combine event, RGB or infrared, IMU and time-of-flight data – intrinsically or at system level – to strengthen perception in degraded and fast-moving environments through complementary sensing.

Ecosystem-ready integration

Compatibility with mainstream robotics and smart-system frameworks, with integration accelerators for embedded Linux and edge AI and support for integrator and OEM workflows.

Edge-scalable performance

A computing-hardware abstraction layer optimized for embedded real-time systems, scaling from ultra-low-power modules to larger systems while preserving low latency and compact deployment.

From event data to application intelligence

Hearth brings together the tools needed to explore event data, develop perception pipelines and prepare applications for integration.

Visualize, record and tune

Use a graphical environment to explore live and recorded event data from supported Event-Based Vision systems.

  • Visualize event streams and provided datasets
  • Control visualization parameters, including accumulation time and fps rate
  • Record and replay event data
  • Export visualizations to AVI video
  • Configure, save and load sensor settings
Hearth interface: streaming view with accumulation time, frame-rate and external trigger monitoring

Develop with C++ and Python

Build Event-Based Vision applications through comprehensive C++ and Python APIs.

  • Run natively on Linux and Windows
  • Connect to Prophesee-compatible vision systems
  • Access extensive documentation and code examples
  • Develop through available training and learning resources
// Count and time-stamp events from an event-based camera
#include <hearth/sdk/driver/camera.h>
#include <hearth/sdk/base/events/event_cd.h>

class EventAnalyzer {
public:
    int global_counter = 0;              // events processed
    Hearth::timestamp global_max_t = 0;  // highest timestamp

    // camera callback: statistics on the received events
    void analyze_events(const Hearth::EventCD *begin,
                        const Hearth::EventCD *end) {
        global_max_t = (end - 1)->t;
        for (const Hearth::EventCD *ev = begin; ev != end; ++ev) {
            ++global_counter;
        }
    }
};

Applications
that grow with the platform

Hearth brings together Prophesee algorithms, AI components and a growing library of application software for building Event-Based Vision systems – from event-data exploration and calibration to machine learning, computer vision and application analytics.

Machine learning
Computer vision
3D computer vision
Analytics
Calibration

Machine learning

Train, evaluate and run models designed for asynchronous event data.

Detection inference

Unlock the potential of Event-Based machine learning, with a set of dedicated tools providing everything you need to start execution of Deep Neural Network (DNN) with events.

Detection training

Train your own Object Detection application with our ready-to use training framework.

Detection KPIs

Evaluate your detection performance with our Object Detection KPI toolkit in line with the latest COCO API.

Optical flow inference

Predict optical flow from Event-Based data leveraging our pretrained Flow Model.

Optical flow training

Train your Optical Flow application with our custom-built FlowNet training framework.

Gesture classification inference

Run a live Rock Paper Scissors game with our pre-trained model on live stream or on event-based recordings.

Gesture classification training

Train your own object classifier with this tailor-made Event-based training pipeline.

Video-to-event

Bridge frame-based and event-based worlds with our Video to Event Pipeline.

Event-to-video

Build grayscale images based on events with this neural network.

Corner detection and tracking

Detect and track corners in an event stream with high efficiency.

Computer vision

Extract motion and visual structure directly from event streams.

Optical flow

Rediscover this fundamental computer vision building block, but with an event twist.

XYT visualization

Discover the power of time – space continuity for your application by visualizing your data with our XYT viewer.

Ultra slow motion

Slow down time, down to the time-resolution equivalent of over 200,000+ frames per second, live.

3D computer vision

Support spatial perception through event-based tracking, active markers and stereo vision.

Edgelet tracking

Track 3D edges and/or Fiducial markers for your AR/VR application.

Active marker

Track active markers at ultra-high speed in real-time.

Stereo matching

Estimate semi-dense depthmaps by matching natural features in synchronized stereo event streams.

Analytics

Turn event data into application-level measurements and operational information.

Particle-size monitoring

Control, count and measure the size of objects moving at very high speed in a channel or a conveyor.

Object tracking

Track moving objects in the field of view. Leverage the low data-rate and sparse information provided by event-based sensors.

Vibration monitoring

Monitor vibration frequencies continuously, remotely, with pixel precision, by tracking the temporal evolution of every pixel in a scene.

Spatter monitoring

Track small particles with spatter-like motion.

High-speed counting

Count objects at unprecedented speeds, high accuracy, generating less data and without any motion blur.

Calibration

Calibrate individual cameras and multi-camera configurations for accurate system integration.

Calibration

Deploy your applications in real-life environments and control all the optical parameters of your event-based systems.

Extrinsic calibration

Combine multiple cameras to enable 3D geometrical reasoning in your application.

Build and deploy with Hearth

Hearth provides a shared software and intelligence foundation across Prophesee products, with the tools and application capabilities in each offer shaped around its purpose.

Embra HD

Record, visualize and explore event data, then develop and evaluate your own Event-Based Vision applications.

Explore Embra HD

Mantara®

Access algorithms and applications developed specifically for ground-based drone detection and tracking.

Explore Mantara

Mantara Lite

Access onboard detection and tracking applications developed for airborne platforms.

Explore Mantara Lite

Nexara

Access application intelligence developed for high-speed industrial inspection and infrastructure monitoring.

Explore Nexara

From Metavision SDK5 PRO to Hearth

Hearth succeeds Metavision SDK5 PRO as a platform for full-stack embedded integration and deployment. Designed as a long-term software foundation, it will evolve to maintain compatibility with current and future Prophesee sensors and help you optimize the performance of your Event-Based Vision application.

Build what comes next with Hearth

Discuss your application, target platform and deployment goals with Prophesee experts.

Inquire