Glossary

This section provides meanings and definitions for terms that one may find in Prophesee documentations.

Generic terms

Term	Definition	Meaning
AER	Address-Event Representation	Transmission protocol for event-based data. An “event” is a data packet containing the pixel’s address in the array plus a payload, e.g. contrast detection polarity
AFK	Anti FlicKer	Technique allowing to remove flickering events (in SW or HW)
ATIS	Asynchronous Time-based Image Sensor	Name for Prophesee’s family of frame-free contrast-triggered pixel-individually auto-sampling image sensors with time-domain intensity encoding
Biases		Set of settings allowing to control Prophesee sensors (e.g. their sensitivity, bandwidth, etc)
CD event	Contrast Detection	Indicates an “event” triggered by temporal contrast (relative intensity change). CD events have a polarity, negative or positive (CD ON or CD OFF), depending on if triggered by decreasing or increasing light intensity
Crazy Pixel		Pixel generating spontaneous background activity; data from crazy pixels can be removed with filtering algorithms (e.g. STC filter)
DSP	Digital Signal Processor	Specialized microprocessor chip, with its architecture optimized for the operational needs of digital signal processing.
EM event	Exposure Measurement	Event-based gray-level data representation using time-domain encoding of the intensity information (can also be found as “APS”)
EPS	Events Per Second	Readout Speed. For example, EPSMAX - Maximum throughput (Events per second, EPS) of the event readout system
ERC	Event Rate Control	Technique used to control the maximum event rate of a sensor
ESP	Event Signal Processing	Specialized microprocessor chip, with its architecture optimized for the operational needs of event signal processing.
Event		Sensor-generated data packet containing the pixel’s address in the array, a payload (e.g. contrast detection polarity), and a timestamp
EVK	Evaluation Kit	Prophesee camera with the capability to visualize, record and process data on a PC
EVT	Event Type Definition	Prophesee event-based Processing Systems gather data from the sensor and encode these data in a proprietary event stream format. This event format has been specified to strike a balance between compactness and robustness.
FOV	Field Of View	Area that can be seen by the camera with its optics at the given position
Latency		Delay of response for a pixel. It is the time between a stimulus light event seen by a CD pixel and the related event’s timestamp. This term should not be confused with the timestamp resolution. Typical values for our last generation of sensor is a latency of 100us and a temporal resolution of 1us.
MIPI	Mobile Industry Processor Interface	Interface supported by some of Prophesee products and sensors
NCT	Nominal Contrast Threshold	Contrast at 10 lux with response probability higher than 0.5
ONBOARD		Prophesee Reference System (deprecated), an embedded solution that combines Prophesee sensor and computer vision products on a single board computer platform
Metavision HAL	Hardware Abstraction Layer	Software module part of the SDK which aims at controlling Prophesee event-based cameras and acquiring data
Metavision Studio		Application with GUI allowing to visualize and record data from event-based sensors and also read already recorded data (e.g. datasets)
Metavision SDK		Prophesee SDK allowing to process event-based data. High-performance algorithms are available via APIs, ready to go to production with event-based Vision applications
Polarity		Property of a CD event that can be negative or positive (CD ON or CD OFF) depending on if triggered by decreasing or increasing light
PSM	Particle Size Measurement	Application aimed at measuring size of particles
ROI	Region of Interest	Part of the sensor still active while other pixels are kept in reset, to focus on specific parts of the scene
SoC	System on Chip	Integrated circuit that takes a single platform and integrates an entire electronic or computer system onto it (CPU with additional features, like HW accelerators)
STC filter	Spacial Temporal Contrast	Data filtering algorithm for removing noisy data based on spatio-temporal consistency (presence of other events in space and time) available in HW and SW
TD event	Temporal Difference	Deprecated name for CD events
Timestamp		Time of event generation by Prophesee sensor, in microsecond precision
Trail Filter		Data filtering algorithm for filtering events at a pixel level where all events following the first event for the same pixel are discarded available in HW and SW)
Trigger		Input or output signals (Trigger In or Out) allowing to control the sensor and synchronize data between several devices
VCEB	VisionCam Event-Based	Imago technologies reference system using Prophesee sensor
XYT view	X-Y-Time view	3D representation of events data in X-Y-Time space

SW-specific terms

Term	Definition	Meaning
Asynchronous algorithm		Algorithm processing time slices of fixed duration and allowing to retrieve an outcome at a fixed refresh rate rather than for each processed buffer of events
IOU	Intersection over Union	Evaluation metric used to measure the accuracy of an object detector on a particular dataset
NMS	Non Maximum Suppression	Class of algorithms used in computer vision to select one entity (e.g. bounding boxes) out of many overlapping entities
Pipeline		Class aimed at simplifying the writing of algorithms; it processes data as a pipeline of processing units (stages) and controls their execution
Pipeline stage		Processing unit (stage) in a pipeline; each processing stage can have previous stages and next stages, thus forming a directed graph
ROS	Robot Operating System	Set of software libraries and tools that help to build robot applications
Sample		Code example aimed at demonstrating a concept as simple as possible, it serves as a tutorial and is delivered as C++ and Python source code. Some of our samples are also delivered as pre-compiled executable (see Tool)
Synchronous algorithm		Algorithm directly processing each received buffer of events and generating an outcome for each processed buffer
Tool		Software tool aimed at solving specific needs (e.g. focusing, calibration, data conversion, etc). Delivered as a sample in source code and also as pre-compiled binaries, ready for execution