Fix HF Space config: enable docker

Dockerfile

@@ -1,5 +1,7 @@
 # ── Base image ──────────────────────────────────────────────────────────────
 # Python 3.11 slim keeps the image lean while matching the dev environment.
+RUN useradd -m -u 1000 user
+USER user
 FROM python:3.11-slim
 
 # ── System dependencies ──────────────────────────────────────────────────────

README.md

@@ -7,72 +7,237 @@ sdk: docker
 pinned: false
 license: mit
 ---
+# StressDetect - MindView
 
-# 🧠 StressDetect
+An end-to-end stress detection platform with:
 
-A full-stack mental health application for real-time stress detection with
-personalized interventions — built as an end-to-end ML systems project.
+- A secure FastAPI backend for analysis, history, and personalization
+- A Streamlit dashboard for check-ins, trends, and interventions
+- A deep learning pipeline for training and retraining text classifiers
+- A safety-first intervention engine with crisis escalation pathways
+
+This repository is designed as both a usable application and a complete ML systems project.
+
+## What This Project Does
+
+Given short or long free-text input, the system:
+
+1. Cleans and processes text
+2. Predicts stress probability with uncertainty-aware inference
+3. Assigns a human-readable stress level
+4. Applies temporal modeling from user history
+5. Returns personalized intervention guidance
+6. Stores encrypted session history and feedback
+
+## Key Features
+
+- Model inference with MC-Dropout uncertainty estimation
+- Adaptive thresholding and temporal stress profiling
+- Rule-based intervention engine with trigger categories
+- Crisis keyword circuit breaker with immediate safety messaging
+- JWT authentication and bcrypt password hashing
+- Encrypted stress history at rest
+- RL-style feedback loop from user and LLM reward signals
 
 ## Architecture
 
+```text
+Streamlit UI (port 7860)
+    |
+    | REST
+    v
+FastAPI Backend (port 8000)
+    |
+    +-- Model Inference (CNN/Transformer)
+    +-- Temporal Model
+    +-- Intervention Engine
+    +-- SQLite Database (users, sessions, feedback)
 ```
-┌─────────────────────────────────────────────────┐
-│                  Streamlit UI                   │  ← port 7860 (public)
-│  Dashboard · History & Analytics · Settings     │
-└──────────────────┬──────────────────────────────┘
-                   │ REST (localhost)
-┌──────────────────▼──────────────────────────────┐
-│               FastAPI Backend                   │  ← port 8000 (internal)
-│  /analyze · /history · /feedback · /personalize │
-└──────────────────┬──────────────────────────────┘
-                   │
-       ┌───────────┼───────────┐
-       ▼           ▼           ▼
-   CNN Model   SQLite DB   Intervention
-  (MC-Dropout) (encrypted)   Engine
+
+Main runtime entry points:
+
+- `app.py`: boots FastAPI in-process and starts Streamlit (useful for Spaces)
+- `api/main.py`: API routes, auth, inference, persistence
+- `ui/app.py`: dashboard UI
+
+## Project Layout
+
+```text
+StressDetection/
+|- api/                  FastAPI app and request handling
+|- checkpoints/          Trained model checkpoint (model.pt)
+|- data/                 Dataset and evaluation files
+|- database/             SQLite access and schema logic
+|- intervention/         Recommendation + temporal modeling
+|- models/               Neural network architectures
+|- scripts/              Data prep and model download utilities
+|- security/             JWT, password hashing, encryption
+|- tests/                Unit and integration tests
+|- training/             Train and retrain scripts
+|- ui/                   Streamlit frontend
+|- utils/                Shared helpers
+|- Dockerfile            Container build
+|- supervisord.conf      Process supervision
+|- requirements.txt      Runtime dependencies
+|- requirements-train.txt Training and testing dependencies
+```
+
+## Quick Start (Local)
+
+### 1. Create Environment and Install
+
+Windows PowerShell:
+
+```powershell
+python -m venv .venv
+.\.venv\Scripts\Activate.ps1
+pip install -r requirements.txt
 ```
 
-## Model
+Linux/macOS:
+
+```bash
+python -m venv .venv
+source .venv/bin/activate
+pip install -r requirements.txt
+```
 
-- **Multichannel 1D CNN** with multi-head self-attention (4 heads)
-- **MC-Dropout ensemble** (3 passes) for uncertainty estimation
-- **FPR-constrained threshold calibration** (max FPR 20%)
-- **Focal loss** + cosine LR warmup + early stopping
-- Checkpoint: [`Ace-119/stress-detection-cnn`](https://huggingface.co/Ace-119/stress-detection-cnn)
+### 2. Download Model Checkpoint
 
-## Safety
+```bash
+python scripts/download_model.py
+```
 
-- **4-layer intervention engine** with 988 crisis circuit breaker
-- Crisis keywords (suicide/self-harm) → immediate 988 lifeline, pipeline halts
-- 8 trigger categories: sleep, work, exam, money, relationship, health, grief, loneliness
-- Escalation tracker: 3+ consecutive high-stress sessions → professional referral
+### 3. Run Backend API
 
-## Features
+```bash
+uvicorn api.main:app --host 0.0.0.0 --port 8000 --reload
+```
 
-| Feature | Detail |
-|---|---|
-| Stress scoring | CNN probability + MC-Dropout uncertainty |
-| Temporal profiling | Adaptive threshold, velocity, volatility |
-| Interventions | Progressive step-by-step guided flow |
-| RL feedback loop | User + LLM-as-judge reward signal |
-| Personalization | Per-user score bias from feedback history |
-| Analytics | Timeline, calendar heatmap, polar chart, trigger frequency |
-| Security | JWT auth, bcrypt passwords, AES-256 history encryption |
+### 4. Run Dashboard
 
-## Quick Start (local)
+In a second terminal:
 
 ```bash
-git clone https://github.com/Ace-119/StressDetection
-cd StressDetection
-pip install -r requirements-train.txt   # full deps including training
-python scripts/download_model.py        # pulls model.pt from HF Hub
-make dev                                # FastAPI :8000 + Streamlit :8501
+streamlit run ui/app.py
 ```
 
-## Crisis Resources
+Open:
+
+- API docs: http://localhost:8000/docs
+- Dashboard: http://localhost:8501
+
+## Makefile Commands
+
+Common workflows:
+
+- `make install` install runtime deps
+- `make download` download checkpoint
+- `make run` run FastAPI
+- `make dashboard` run Streamlit
+- `make preprocess` prepare dataset
+- `make train` train model
+- `make retrain` retrain with feedback
+- `make test` run tests
+- `make lint` run Ruff lint fixes
+
+Note: the file is named `MakeFile` in this repo.
+
+## API Endpoints
+
+Public:
+
+- `GET /health` service heartbeat
+- `GET /model/info` model metadata
+- `POST /register` create account and return token
+- `POST /login` authenticate and return token
+
+Authenticated (Bearer JWT):
+
+- `POST /analyze` run stress analysis + interventions
+- `GET /history` fetch previous sessions
+- `POST /feedback` submit user feedback for replay/reward
+- `GET /feedback/stats` feedback summary
+- `GET /personalization` personalization status
+
+## Training Workflow
+
+### 1. Prepare Data
+
+```bash
+python scripts/data_preprocessing.py
+```
+
+Expected core training file:
+
+- `data/processed/unified_stress.csv`
+
+### 2. Train Model
+
+```bash
+python training/train.py \
+  --model cnn \
+  --epochs 15 \
+  --batch-size 64 \
+  --lr 1e-3 \
+  --data data/processed/unified_stress.csv \
+  --eval-set data/eval/happy_neutral_eval.csv \
+  --output checkpoints/model.pt
+```
+
+Supported model options include `cnn`, `deberta`, and `minilm`.
+
+### 3. Retrain from Feedback (Optional)
+
+```bash
+python training/retrain.py
+```
+
+## Testing
+
+Run the full test suite:
+
+```bash
+pytest tests/ -v
+```
+
+Or with make:
+
+```bash
+make test
+```
+
+## Security and Safety
+
+Security:
+
+- Passwords hashed with bcrypt
+- JWT-based authenticated access
+- Encrypted session history in SQLite-backed storage
+
+Safety behavior:
+
+- Crisis phrase detection triggers immediate escalation messaging
+- Multi-level intervention outputs aligned to stress severity
+- Temporal escalation tracking across sessions
+
+## Docker and Deployment
+
+Containerized deployment is available through `Dockerfile`.
+
+For single-process hosting environments, `app.py` starts FastAPI and Streamlit together.
+
+## Troubleshooting
+
+- API cannot connect from UI:
+  - Ensure backend is running on port 8000
+  - Set `API_URL` if backend is remote
+- Missing checkpoint:
+  - Run `python scripts/download_model.py`
+- Import or dependency errors:
+  - Reinstall with `pip install -r requirements.txt`
+  - For training extras use `requirements-train.txt`
 
-This app always surfaces crisis resources when needed.
+## License and Use
 
-**988 Suicide & Crisis Lifeline** — Call or text **988** (US)  
-**Crisis Text Line** — Text HOME to **741741**  
-**SAMHSA Helpline** — 1-800-662-4357 (free, confidential, 24/7)
+This project is intended for research and educational use. It is not a medical device and does not replace professional diagnosis or emergency care.