shm-detect-tool 0.8.2

Shmoo log analysis tool based on PyTorch CNN

SHM Detect Tool — Shmoo Log Analysis with CNN

Automatically classify shmoo plot pass/fail patterns using a PyTorch CNN

---

Table of Contents

• Overview
• Features
• Installation
• Usage
  • CLI Usage
  • GUI Usage
  • Web App Usage
• Screenshots
• License

---

Overview

The SHM Detect Tool uses a CNN (Convolutional Neural Network) built on PyTorch to automatically analyse shmoo plot datalogs from ATE (Automated Test Equipment). It parses raw shmoo text logs, converts them into image-like tensors, and classifies each plot's pass/fail pattern — replacing tedious manual inspection with fast, repeatable AI-based detection.

Background

• Challenge: Engineers spend significant time visually inspecting shmoo plots to identify pass/fail patterns, margins, and anomalies across multiple test sites and conditions.
• Solution: SHM Detect Tool automates this process by training an AlexNet-based CNN to classify shmoo patterns, generating structured XLSX reports with pass/fail highlighting and site-to-site correlation.

---

Features

• CNN-Based Shmoo Classification — AlexNet architecture trained on labeled shmoo datasets
• Shmoo Log Parsing — convert raw text shmoo logs to structured CSV with configurable keyword patterns
• XLSX Report Generation — automated reports with pass/fail highlighting per shmoo plot
• Site-to-Site Correlation — compare shmoo results across multiple test sites with overlay reports
• Parallel Plot Support — side-by-side multi-site shmoo comparison in XLSX
• CLI Interface — scriptable commands for training, analysis, and correlation
• GUI Interface — PyQt5-based desktop application for interactive workflows
• Web App Interface — Streamlit-based web UI for browser-based analysis
• Configurable Parsing — JSON config files to support different shmoo log formats (S2S, CHAR Studio, etc.)

---

Installation

Requirements

• Python 3.9 – 3.14
• Windows, macOS, or Linux

Install from PyPI (Recommended)

# Basic installation (CLI + core analysis)
pip install shm-detect-tool

# With GUI support (PyQt5)
pip install shm-detect-tool[gui]

# With web app support (Streamlit)
pip install shm-detect-tool[webapp]

After installation, the shm-detect command is immediately available:

shm-detect --version
shm-detect --help

Install from Source

git clone https://github.com/showjim/SHM_Detect_Tool.git
cd SHM_Detect_Tool
python -m venv venv
venv\Scripts\activate        # Windows
# source venv/bin/activate   # macOS/Linux
pip install -e .[gui,webapp]

---

Usage

CLI Usage

# Analyse a shmoo log and generate an XLSX report
shm-detect analyse --log shmoo.txt --config SHM_keywords_setting.json

# Train a new CNN model
shm-detect train --dataset custom_SHM_data.csv

# Compare shmoo results across multiple log files
shm-detect correlate --files file1.txt file2.txt --config SHM_keywords_setting.json

# Show version
shm-detect -V

See doc/CLI_USAGE.md for the full CLI reference including all options and examples.

GUI Usage

Launch the desktop application:

python SHM_Detect_Tool.py

Workflow

1. Convert Shmoo — load a raw shmoo .txt log and convert it to CSV
2. Detect Shmoo — select the CSV file, run CNN inference, and review the generated XLSX report
3. Train CNN — (optional) train a new model from a labeled dataset

Web App Usage

Launch the Streamlit-based web interface:

streamlit run SHM_Detect_Tool_webapp.py

---

Screenshots

---

License

This project is licensed under the GNU General Public License v3.0.

---

Built for the semiconductor ATE community