🔮 ISP Defect Predictor

AI-Powered Defect Prediction for Camera & ISP Quality Programs

Analyzes historical defect data → Predicts risk areas → Prevents escapes before they happen

📖 How It Works · 🚀 Quick Start · 📊 Features

---

📌 What Is This?

The ISP Defect Predictor is an AI-powered quality intelligence tool built for Samsung camera and ISP (Image Signal Processing) programs. It ingests historical defect data from previous Galaxy flagship releases and predicts where defects are most likely to occur in the current or future program.

Built from 13 consecutive Samsung Galaxy flagship releases (S22 → S26) with zero S1 escapes.

---

🎯 Problem It Solves

Without This Tool	With This Tool
Defects found late in cycle	Risks identified at sprint start
Same components fail repeatedly	Pattern recognition prevents repeat failures
Manual triage of 500+ test cases	AI-ranked test priority list
Reactive QA (find & fix)	Predictive QA (prevent & protect)

---

⚙️ How It Works

Historical Defect Data (CSV/JSON)
         ↓
   Data Analysis Engine
   ─ Defect density per component
   ─ Severity distribution (S1/S2/S3/S4)
   ─ Repeat failure patterns
   ─ Sprint-over-sprint trend
         ↓
   AI Prediction Model
   ─ Risk score per component (0-100)
   ─ Predicted defect count
   ─ High-risk test areas flagged
         ↓
   Output Report
   ─ Risk heatmap by component
   ─ Top 10 high-risk areas
   ─ Recommended test coverage boost
   ─ Executive summary for Director review

---

📊 Features

• 📈 Defect Trend Analysis — Sprint-wise defect velocity tracking
• 🎯 Component Risk Scoring — ISP pipeline components ranked by risk (0–100)
• 🔁 Repeat Failure Detection — Identifies components with recurring defects
• 📋 Test Prioritization — Recommends which test suites to run first
• 📊 Severity Distribution — S1/S2/S3/S4 breakdown per component
• 🤖 AI Root Cause Clustering — Groups similar defects automatically
• 📄 Executive Report — Director-level summary with actionable recommendations

---

🏗️ Architecture

isp_defect_predictor/
│
├── data/
│   ├── historical_defects.csv    ← Input: past defect data
│   └── component_map.json        ← ISP component definitions
│
├── engine/
│   ├── analyzer.py               ← Defect pattern analysis
│   ├── predictor.py              ← Risk score calculation
│   └── reporter.py               ← Report generation
│
├── output/
│   ├── risk_heatmap.html         ← Visual risk map
│   └── executive_report.pdf      ← Director-level summary
│
├── tests/
│   └── test_predictor.py         ← Unit tests
│
├── requirements.txt
└── README.md

---

🚀 Quick Start

# Clone
git clone https://github.com/vanichalla24/isp_defect_predictor.git
cd isp_defect_predictor

# Install
pip install -r requirements.txt

# Add your defect data
cp your_defects.csv data/historical_defects.csv

# Run prediction
python engine/predictor.py

# View report
open output/risk_heatmap.html

---

📥 Input Format

defect_id,component,severity,sprint,description,status
DEF-001,Camera_HAL,S1,Sprint_42,AF fails in low light,Closed
DEF-002,ISP_Pipeline,S2,Sprint_42,Color shift in Pro mode,Closed
DEF-003,Night_Mode,S1,Sprint_43,Overexposure on main cam,Open

---

📤 Output Sample

🔴 HIGH RISK (Score 85/100)  → Camera HAL3 Layer
🔴 HIGH RISK (Score 78/100)  → ISP Night Mode Pipeline
🟡 MEDIUM RISK (Score 54/100) → 3A Algorithm (AE/AWB/AF)
🟢 LOW RISK (Score 22/100)   → Gallery Rendering

---

👩‍💻 Author

Vaani Challa — QA Architect | Samsung SRIB | 17+ Years 13 Galaxy flagship releases · Zero S1 escapes · Dual Best Award (S25 + S26)

---

📄 License

MIT © Vaani Challa