Event Camera Hardware: Sensors, Evaluation Kits and Platforms
Hardware Overview
Several companies and research groups are actively developing event-based vision sensors and cameras, enabling applications in robotics, industrial automation, mobile devices, and scientific research. Modern event-based sensors are typically implemented using Dynamic Vision Sensor (DVS) architectures or Hybrid Vision Sensors (HVS) that combine event sensing with conventional frame capture. These technologies allow cameras to detect brightness changes with microsecond temporal precision while maintaining high dynamic range and low power consumption.
Available Sensors and Vendors
Major Companies Working on Event-Based Vision
Prophesee
Prophesee is one of the leading companies in commercial event-based vision technology. Its sensors and cameras are widely used in industrial and research applications. The company also provides the Metavision® Intelligence Suite, a comprehensive software ecosystem for event-based vision development. Examples include sensors co-developed with Sony such as IMX636 and IMX637.
Sony
Sony has developed stacked event-based vision sensors integrated with Prophesee technology. Sensors such as IMX636/637 feature small pixel sizes and high event throughput, targeting industrial, robotics, and machine vision applications.
iniVation
iniVation develops neuromorphic vision sensors and event-based cameras including the DAVIS and DVXplorer families. Their devices are widely used in research and education, and they provide developer tools and documentation for event-driven vision systems.
Samsung
Samsung has developed several Dynamic Vision Sensor (DVS) designs as part of its research in event-based vision. Most of this work has been presented in academic publications and prototype demonstrations, focusing on high-speed sensing and low-latency event processing.
As an example of an application of Samsung's DVS technology, SmartThings Vision is a smart home monitoring device that detects motion using an event-based sensor instead of continuously recording video.
Evaluation Kits and Camera Modules
Prophesee
Prophesee offers several evaluation kits designed for experimenting with event-based vision and developing prototype applications. These kits allow developers to capture event data and build end-to-end embedded vision pipelines closer to production environments.
Prophesee evaluation hardware is typically organized into two main categories:
- USB evaluation cameras for PC-based experimentation.
- Embedded starter kits for developing event-based applications on embedded platforms.
| Evaluation hardware | Sensor | Host platform | Purpose |
|---|---|---|---|
| EVK4 HD | IMX636 | PC via USB | High-resolution event-camera evaluation |
| EVK3 GENX320 | GENX320 | PC via USB | GENX320 sensor evaluation |
| Raspberry Pi 5 starter kit | GENX320 | Raspberry Pi 5 | Embedded event-vision prototyping |
| AMD Kria KV260 starter kit | IMX636 or GENX320 | AMD Kria KV260, MIPI sensor interface | FPGA-based event-vision development |
| STM32F7 starter kit | GENX320 | STM32F746G-DISCO, parallel DCMI sensor interface | Microcontroller-based event-vision development |
| EVK3D structured-light system | IMX636 | Onboard FPGA processing; USB 3.0 output to host PC | Active light triangulation with 3D point-cloud output |
Sources: Prophesee evaluation kits, embedded interface and sensor compatibility, STM32F7 product brief, and EVK3D product brief.
The sensors mentioned in the table have the following characteristics:
- IMX636 (1280 × 720): a high-resolution event sensor developed in collaboration between Sony and Prophesee
- GENX320 (320 × 320): a compact and power-efficient event sensor designed for embedded and edge applications
iniVation
iniVation provides several event-based cameras designed for development and research. These cameras are typically available as ready-to-use devices with USB connectivity and developer tooling to facilitate experimentation with event-based vision systems.
| Camera | Resolution (pixels) | Output | Maximum event throughput | Distinguishing features |
|---|---|---|---|---|
| DVXplorer | 640 × 480 | Events | 165 million events/s | Aluminium housing, CS-mount lens, multi-camera synchronization |
| DVXplorer Lite | 320 × 240 | Events | 100 million events/s | Lower-resolution model, plastic housing, CS-mount lens, multi-camera synchronization |
| DVXplorer Micro | 640 × 480 | Events | 450 million events/s | Compact 24 × 27.5 × 29.7 mm housing, S-mount lens; no multi-camera synchronization |
| DAVIS346 | 346 × 260 | Events and conventional frames | 12 million events/s | Concurrent event and frame output; frames up to 40 fps; multi-camera synchronization |
DVXplorer cameras generate event streams using DVS sensors, while DAVIS cameras combine event output with conventional frame-based imaging.
Supported Platforms
Event-based vision systems require computing platforms capable of handling high-rate event streams and running specialized processing algorithms. Depending on the application, these platforms may range from embedded devices to high-performance systems.
NVIDIA Jetson
- Jetson Orin NX: Great option for real-time and high-performance applications. Suitable for complex pipelines and AI workloads.
For a practical integration example, follow the GENX320 setup guide for NVIDIA Jetson Orin NX.
Raspberry Pi
- Raspberry Pi 5: Compact and low-cost platform for experimentation and integration. Suitable for basic pipelines, but not ideal for high-throughput processing. Some vendors, such as Prophesee, offer evaluation kits where event sensors connect via MIPI CSI-2, enabling compact embedded setups.
For a compact embedded setup, follow the GENX320 setup guide for Raspberry Pi 5.