How relativity1395/evnt_work works
This repository is event-camera software for ingesting DVXplorer events, detecting and visualizing features, and developing event-based optical-flow and tracking/state-estimation algorithms. The visible driver opens the camera, reads event packets, uses the fast detector, and renders detected features; README also describes an optical-flow-filter visualization. The tree identifies additional algorithm stages, but their source was not sampled, so links among them are omitted rather than inferred. The sampled EVIO entry point is only a stub.
Main entry points: DVX_Drivers.cpp DVX event driver · dvxplore_simple.cpp Flow visualization · Types.hpp Event and IMU types · fast_detector.cpp Fast detector
Components 16
Capture and visualization
- DVX event driver
src/DVX_Drivers.cpp - Event and IMU types
include/Types.hpp - Fast detector
third_party/fast_detector.cpp - Feature display
src/DVX_Drivers.cpp - Flow visualization
src/dvxplore_simple.cpp
Event vision algorithms
- Optical flow
src/OpticalFlow.cpp - Template alignment
src/TemplateAlignment.cpp - Outlier rejection
src/OutlierRejection.cpp - Temporal window
src/TemporalWindow.cpp - Feature state
src/Feature.cpp - Landmark weighting
src/WeightDist.cpp
Tracking and estimation
- Feature tracker
src/FeatureTracker.cpp - State estimator
src/StateEstimator.cpp - EVIO core
src/EVIO.cpp
Other components
- Camera operator
- DVXplorer camera
Connections 9
- Camera operator → DVX event driver: starts capture
- DVX event driver → DVXplorer camera: opens and reads
- DVXplorer camera → DVX event driver: sends packets
- DVX event driver → Event and IMU types: creates events
- DVX event driver → Fast detector: tests features
- DVX event driver → Feature display: renders features
- Camera operator → Flow visualization: runs visualization
- Flow visualization → DVXplorer camera: opens and reads
- DVXplorer camera → Flow visualization: sends packets