# hemal-eng/sanket architecture

> Architecture diagram of the GitHub repository hemal-eng/sanket, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-11-22). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/hemal-eng/sanket
- Repository: https://github.com/hemal-eng/sanket

## Overview

SANKET is an offline, wrist-worn Bangla air-writing recognizer. The primary workflow captures IMU motion, detects a gesture, runs on-device INT8 inference, then displays and speaks confident results. Python tooling supports data collection, model training/testing, and audio preparation. The vendored BanglaText library supplies rendering; its font-conversion utilities form a separate supporting branch. Some relationships rely on README descriptions because the corresponding implementation was not sampled.

Main entry points: [`firmware_infer/src/main.cpp`](https://github.com/hemal-eng/sanket/blob/main/firmware_infer/src/main.cpp) (Gesture Firmware), [`references/BanglaText-main/extras/BanglaFontConverter/BanglaFontConverter/main.cpp`](https://github.com/hemal-eng/sanket/blob/main/references/BanglaText-main/extras/BanglaFontConverter/BanglaFontConverter/main.cpp) (Converter CLI), [`python/train.py`](https://github.com/hemal-eng/sanket/blob/main/python/train.py) (Training Pipeline), [`references/BanglaText-main/src/Utility/BanglaTextRenderer.cpp`](https://github.com/hemal-eng/sanket/blob/main/references/BanglaText-main/src/Utility/BanglaTextRenderer.cpp) (BanglaText Renderer)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_capture["Capture &amp; Data"]
  node_logger["Data Logger<br/>[main.cpp]"]
  node_collector["Collection GUI<br/>[collector.py]"]
end

subgraph group_model["Model Tooling"]
  node_training["Training Pipeline<br/>[train.py]"]
  node_hosttest["Host Test GUI<br/>[infer_gui.py]"]
  node_audio_prep["Audio Preparation<br/>[prepare_audio.py]"]
end

subgraph group_device["Wearable Runtime"]
  node_inference["Gesture Firmware<br/>[main.cpp]"]
  node_model_data["INT8 Model<br/>[model_data.h]"]
  node_normalization["Normalization &amp; Classes<br/>[model_norm.h]"]
  node_audio["DFPlayer Audio<br/>[audio.cpp]"]
  node_font["Embedded Font"]
end

subgraph group_language["Bangla Rendering"]
  node_renderer["BanglaText Renderer"]
  node_glyphdecoder["Glyph Decoder<br/>[BTGlyphDecoder.cpp]"]
end

subgraph group_fonttools["Font Utilities"]
  node_font_extraction["Font Extraction"]
  node_font_generation["Header Generation"]
  node_font_converter["Converter CLI<br/>[main.cpp]"]
  node_font_headers["Font Header Assets"]
end

node_wearer(("Wrist User"))
node_imu["MPU6050 IMU"]
node_speaker["Speaker"]
node_oled["OLED Display"]

node_wearer -->|"moves wrist"| node_imu
node_imu -.->|"streams samples"| node_logger
node_logger -.->|"sends recordings"| node_collector
node_collector -->|"provides samples"| node_training
node_training -->|"exports model"| node_model_data
node_wearer -->|"writes gesture"| node_imu
node_imu -->|"supplies motion"| node_inference
node_inference -->|"runs inference"| node_model_data
node_inference -->|"reads parameters"| node_normalization
node_inference -->|"shows word"| node_oled
node_inference -->|"renders Bangla"| node_renderer
node_renderer -.->|"uses glyph data"| node_font
node_renderer -->|"decodes glyphs"| node_glyphdecoder
node_inference -->|"requests playback"| node_audio
node_audio -->|"plays recording"| node_speaker
node_audio_prep -.->|"prepares audio"| node_speaker
node_training -.->|"supports testing"| node_hosttest
node_font_converter -->|"extracts font data"| node_font_extraction
node_font_converter -->|"generates headers"| node_font_generation
node_font_converter -->|"loads font assets"| node_font_headers

click node_logger "https://github.com/hemal-eng/sanket/blob/main/firmware/src/main.cpp"
click node_collector "https://github.com/hemal-eng/sanket/blob/main/python/collector.py"
click node_training "https://github.com/hemal-eng/sanket/blob/main/python/train.py"
click node_hosttest "https://github.com/hemal-eng/sanket/blob/main/python/infer_gui.py"
click node_audio_prep "https://github.com/hemal-eng/sanket/blob/main/python/prepare_audio.py"
click node_inference "https://github.com/hemal-eng/sanket/blob/main/firmware_infer/src/main.cpp"
click node_model_data "https://github.com/hemal-eng/sanket/blob/main/firmware_infer/src/model_data.h"
click node_normalization "https://github.com/hemal-eng/sanket/blob/main/firmware_infer/src/model_norm.h"
click node_audio "https://github.com/hemal-eng/sanket/blob/main/firmware_infer/src/audio.cpp"
click node_font "https://github.com/hemal-eng/sanket/blob/main/firmware_infer/src/BanglaMNRegular_25pt.h"
click node_renderer "https://github.com/hemal-eng/sanket/blob/main/references/BanglaText-main/src/Utility/BanglaTextRenderer.cpp"
click node_glyphdecoder "https://github.com/hemal-eng/sanket/blob/main/references/BanglaText-main/src/Utility/BTGlyphDecoder.cpp"
click node_font_extraction "https://github.com/hemal-eng/sanket/tree/main/references/BanglaText-main/extras/BanglaFontConverter/BanglaFontConverter/FontDataExtractor"
click node_font_generation "https://github.com/hemal-eng/sanket/tree/main/references/BanglaText-main/extras/BanglaFontConverter/BanglaFontConverter/FontHeaderGeneration"
click node_font_converter "https://github.com/hemal-eng/sanket/blob/main/references/BanglaText-main/extras/BanglaFontConverter/BanglaFontConverter/main.cpp"
click node_font_headers "https://github.com/hemal-eng/sanket/tree/main/references/BanglaText-main/extras/FontsHeader"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_logger,node_collector,node_wearer toneBlue
class node_training,node_hosttest,node_audio_prep toneAmber
class node_inference,node_model_data,node_normalization,node_audio,node_font toneMint
class node_renderer,node_glyphdecoder toneRose
class node_font_extraction,node_font_generation,node_font_converter,node_font_headers,node_imu,node_speaker,node_oled toneIndigo
```

## Components

### Capture & Data

- [Data Logger](https://github.com/hemal-eng/sanket/blob/main/firmware/src/main.cpp): `firmware/src/main.cpp`
- [Collection GUI](https://github.com/hemal-eng/sanket/blob/main/python/collector.py): `python/collector.py`

### Model Tooling

- [Training Pipeline](https://github.com/hemal-eng/sanket/blob/main/python/train.py): `python/train.py`
- [Host Test GUI](https://github.com/hemal-eng/sanket/blob/main/python/infer_gui.py): `python/infer_gui.py`
- [Audio Preparation](https://github.com/hemal-eng/sanket/blob/main/python/prepare_audio.py): `python/prepare_audio.py`

### Wearable Runtime

- [Gesture Firmware](https://github.com/hemal-eng/sanket/blob/main/firmware_infer/src/main.cpp): `firmware_infer/src/main.cpp`
- [INT8 Model](https://github.com/hemal-eng/sanket/blob/main/firmware_infer/src/model_data.h): `firmware_infer/src/model_data.h`
- [Normalization & Classes](https://github.com/hemal-eng/sanket/blob/main/firmware_infer/src/model_norm.h): `firmware_infer/src/model_norm.h`
- [DFPlayer Audio](https://github.com/hemal-eng/sanket/blob/main/firmware_infer/src/audio.cpp): `firmware_infer/src/audio.cpp`
- [Embedded Font](https://github.com/hemal-eng/sanket/blob/main/firmware_infer/src/BanglaMNRegular_25pt.h): `firmware_infer/src/BanglaMNRegular_25pt.h`

### Bangla Rendering

- [BanglaText Renderer](https://github.com/hemal-eng/sanket/blob/main/references/BanglaText-main/src/Utility/BanglaTextRenderer.cpp): `references/BanglaText-main/src/Utility/BanglaTextRenderer.cpp`
- [Glyph Decoder](https://github.com/hemal-eng/sanket/blob/main/references/BanglaText-main/src/Utility/BTGlyphDecoder.cpp): `references/BanglaText-main/src/Utility/BTGlyphDecoder.cpp`

### Font Utilities

- [Font Extraction](https://github.com/hemal-eng/sanket/tree/main/references/BanglaText-main/extras/BanglaFontConverter/BanglaFontConverter/FontDataExtractor): `references/BanglaText-main/extras/BanglaFontConverter/BanglaFontConverter/FontDataExtractor`
- [Header Generation](https://github.com/hemal-eng/sanket/tree/main/references/BanglaText-main/extras/BanglaFontConverter/BanglaFontConverter/FontHeaderGeneration): `references/BanglaText-main/extras/BanglaFontConverter/BanglaFontConverter/FontHeaderGeneration`
- [Converter CLI](https://github.com/hemal-eng/sanket/blob/main/references/BanglaText-main/extras/BanglaFontConverter/BanglaFontConverter/main.cpp): `references/BanglaText-main/extras/BanglaFontConverter/BanglaFontConverter/main.cpp`
- [Font Header Assets](https://github.com/hemal-eng/sanket/tree/main/references/BanglaText-main/extras/FontsHeader): `references/BanglaText-main/extras/FontsHeader`

### Other components

- **Wrist User**
- **MPU6050 IMU**
- **Speaker**
- **OLED Display**

## Connections

- Wrist User → MPU6050 IMU: moves wrist (evidence: `README.md`)
- MPU6050 IMU → Data Logger: streams samples
- Data Logger → Collection GUI: sends recordings
- Collection GUI → Training Pipeline: provides samples (evidence: `README.md`)
- Training Pipeline → INT8 Model: exports model (evidence: `README.md`)
- Wrist User → MPU6050 IMU: writes gesture (evidence: `README.md`)
- MPU6050 IMU → Gesture Firmware: supplies motion (evidence: `README.md`)
- Gesture Firmware → INT8 Model: runs inference (evidence: `firmware_infer/src/main.cpp`)
- Gesture Firmware → Normalization & Classes: reads parameters (evidence: `firmware_infer/src/main.cpp`)
- Gesture Firmware → OLED Display: shows word (evidence: `README.md`)
- Gesture Firmware → BanglaText Renderer: renders Bangla (evidence: `README.md`)
- BanglaText Renderer → Embedded Font: uses glyph data
- BanglaText Renderer → Glyph Decoder: decodes glyphs (evidence: `references/BanglaText-main/src/Utility/BanglaTextRenderer.cpp`)
- Gesture Firmware → DFPlayer Audio: requests playback (evidence: `README.md`)
- DFPlayer Audio → Speaker: plays recording (evidence: `README.md`)
- Audio Preparation → Speaker: prepares audio (evidence: `README.md`)
- Training Pipeline → Host Test GUI: supports testing (evidence: `README.md`)
- Converter CLI → Font Extraction: extracts font data (evidence: `references/BanglaText-main/extras/BanglaFontConverter/BanglaFontConverter/main.cpp`)
- Converter CLI → Header Generation: generates headers (evidence: `references/BanglaText-main/extras/BanglaFontConverter/BanglaFontConverter/main.cpp`)
- Converter CLI → Font Header Assets: loads font assets (evidence: `references/BanglaText-main/extras/BanglaFontConverter/BanglaFontConverter/main.cpp`)
