@rajeev02/edge-ai 0.2.0 → 0.2.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +23 -0
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -16,6 +16,29 @@ Part of [Rajeev SDK](https://github.com/Rajeev02/rajeev-sdk) — cross-platform
|
|
|
16
16
|
- **GPU acceleration** — Optional GPU backend for TFLite and CoreML models
|
|
17
17
|
- **Privacy-first** — All processing happens on device. No data sent to cloud servers.
|
|
18
18
|
|
|
19
|
+
## ⚠️ Important: ML Runtimes & OCR Engine Required
|
|
20
|
+
|
|
21
|
+
This library provides **ML pipeline management, text parsing utilities, and voice intent matching**. It does **NOT** include ML models, OCR engines, or speech recognition.
|
|
22
|
+
|
|
23
|
+
| What the library does | What YOU must provide |
|
|
24
|
+
| ------------------------------------------------- | --------------------------------------------------- |
|
|
25
|
+
| Parse OCR text to extract PAN/Aadhaar/DL numbers | OCR engine to convert images → text first |
|
|
26
|
+
| Validate Aadhaar (Verhoeff checksum), PAN format | Nothing — works standalone |
|
|
27
|
+
| Model lifecycle management (register/load/unload) | TFLite, CoreML, or ONNX runtime + model files |
|
|
28
|
+
| Rule-based voice intent parsing (11 languages) | Speech-to-text engine to convert audio → text first |
|
|
29
|
+
|
|
30
|
+
**OCR functions** (`extractPanNumber`, `extractAadhaarNumber`, `detectDocumentType`) accept **pre-extracted text strings** — you must first convert images to text using an OCR engine:
|
|
31
|
+
|
|
32
|
+
| Platform | Recommended OCR engine |
|
|
33
|
+
| -------------- | ---------------------------------------------------------------------------------------------- |
|
|
34
|
+
| iOS | [Apple Vision Framework](https://developer.apple.com/documentation/vision) |
|
|
35
|
+
| Android | [Google ML Kit Text Recognition](https://developers.google.com/ml-kit/vision/text-recognition) |
|
|
36
|
+
| Cross-platform | [Tesseract.js](https://tesseract.projectnaptha.com/) (web/WASM) |
|
|
37
|
+
|
|
38
|
+
**ML pipeline** manages model state (register → load → ready → unload) but actual inference must be implemented via native ML runtimes on each platform.
|
|
39
|
+
|
|
40
|
+
**Voice AI** parses text commands using regex-based intent detection — it does not record or transcribe audio. Feed it text from a speech-to-text engine like [Google Speech-to-Text](https://cloud.google.com/speech-to-text) or [Whisper](https://openai.com/research/whisper).
|
|
41
|
+
|
|
19
42
|
## Platform Support
|
|
20
43
|
|
|
21
44
|
| Platform | Engine | Status |
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@rajeev02/edge-ai",
|
|
3
|
-
"version": "0.2.
|
|
3
|
+
"version": "0.2.1",
|
|
4
4
|
"description": "Edge AI / On-Device ML Toolkit — OCR, face detection, voice commands, document scanner, smart crop",
|
|
5
5
|
"main": "lib/index.js",
|
|
6
6
|
"author": "Rajeev Kumar Joshi <rajeevjoshi91@gmail.com> (https://rajeev02.github.io)",
|