captainaai/grpcclientapp

How captainaai/grpcclientapp works

VisionClientApp is a Qt desktop camera-and-object-detection client. The principal workflow is: select a Triton server and ready model, capture a camera frame, prepare it for the model, run remote inference, post-process detections, and display the result. The sampled code shows the Qt/Python/OpenCV integration; some implementation details are omitted, so only documented and visible call relationships are drawn.

Main entry points: mainvision.cpp Main window · mainvision.ui Qt interface · main.cpp Application startup

Components 17

Desktop application

Qt interface mainvision.ui
Application startup main.cpp
Main window mainvision.cpp

Vision workflow

Camera capture mainvision.cpp
Frame preparation mainvision.cpp
Frame display mainvision.cpp
Detection post-processing mainvision.cpp
Detection result mainvision.cpp

Remote inference

Embedded Python mainvision.cpp
Server connection mainvision.cpp
Model selection mainvision.cpp
Triton client mainvision.cpp
Model repository mainvision.cpp
Inference request mainvision.cpp

Other components

User
Camera device
Triton server

Connections 25

User → Qt interface: operates
Application startup → Main window: creates
Main window → Qt interface: sets up
Main window → Embedded Python: initializes
Main window → Camera capture: opens
Camera capture → Camera device: reads frames
Main window → Camera capture: updates
Main window → Server connection: handles selection
Server connection → Embedded Python: calls connect
Embedded Python → Triton client: uses
Triton client → Triton server: connects
Embedded Python → Model repository: queries
Model repository → Triton server: reads model status
Main window → Model selection: handles selection
Model selection → Embedded Python: calls select_model
Embedded Python → Triton server: checks readiness and metadata
Main window → Frame preparation: prepares captured frame
Frame preparation → Inference request: passes model input
Main window → Inference request: calls runInference
Inference request → Embedded Python: calls infer
Embedded Python → Triton server: sends inference
Embedded Python → Detection post-processing: decodes predictions
Detection post-processing → Detection result: produces detections
Main window → Frame display: shows frames and results
Frame display → Qt interface: updates viewer

More C++ diagrams: hobuinc/untwine · libnifalcon/libnifalcon · reefwing-software/reefwing-ahrs · tridao/flash-attention-wheels · valkmjolnir/nasal-interpreter

Drawn by GitDiagram on September 24, 2026 from the repository's file tree, README and source files. Read as Markdown ·