noackjona-hash/jonanoackignite

Purpose and entry points

IGNITE is a desktop medical thermal-image analysis client for ThermoAI Vision. Its primary executable entry point is main.py, which prioritizes immediate UI startup while deferring heavy imaging/ML imports. The desktop package is built for distribution with PyInstaller (Ignite.spec) and an installer definition (ignite_installer.iss). A separate static/Vite-style product landing site lives under ignite-landing/src/main.js.

Application bootstrap and responsive UI

The startup path coordinates a lightweight splash/loading experience, background initialization, and the main CustomTkinter event loop. The main dashboard and user-driven analysis controls are centered in:

• main.py

• gui/main_window.py

• gui/theme.py

This is a desktop GUI boundary: UI work must remain on the Tk/CustomTkinter thread, while expensive imports and image analysis should execute asynchronously to avoid blocking interaction.

Thermal image-processing pipeline

The computer-vision layer performs local processing of thermal imagery, including bilateral filtering, CLAHE enhancement, contour segmentation, ROI/foot isolation, and bilateral temperature comparison. Its core processing boundary is:

• image_processing.py

• utils.py

• config.py

The pipeline consumes selected thermal images from the UI, produces derived image/measurement results for presentation, and is intended to keep calculations in memory. OpenCV, NumPy, Pillow, and potentially PyTorch are material runtime dependencies.

Native/Rust computation core

The repository includes a Rust crate alongside the Python application, suggesting selected performance-sensitive or correctness-critical functionality is implemented or being migrated into a native core:

• Cargo.toml

• src/lib.rs

• Cargo.lock

Python/Rust behavioral alignment is explicitly represented by parity-oriented tests, making this an important interface boundary: Python-side processing and the Rust library should preserve equivalent results and data semantics.

Data persistence and audit boundary

Application-local state, persisted settings, and audit-related behavior are separated from visual and processing code:

• storage.py

• audit_log.py

• settings.json

This layer is the likely boundary for retaining user preferences, derived metadata, and compliance-relevant audit records. Raw medical imagery should not be treated as durable application data unless explicitly required, consistent with the stated privacy-first design.

Privacy and analysis-service integration

The intended deployment model supports local image processing and potentially encrypted communication with a local FastAPI-based ThermoAI analysis backend. The desktop client is therefore responsible for controlling the transition between local UI/image handling and any backend request path. Relevant client-side orchestration and settings surfaces are:

• main.py

• config.py

• storage.py

No backend service source is present in this tree; FastAPI/TLS infrastructure is an external integration rather than a repository-owned server runtime.

Packaging, release, and public distribution

Desktop packaging combines Python dependencies with application assets into installable artifacts, while CI builds and releases wheels:

• Ignite.spec

• .github/workflows/build_wheels.yml

• .github/workflows/release.yml

The separate web landing project distributes installation material and marketing assets, including a public Windows installer:

• ignite-landing/src/main.js

• ignite-landing/public/IGNITE_Setup_v1.0.0.exe