I have built the computer vision models in 3 different ways addressing different personas, because not all companies will have a resolute data science team. quality-control manufacturing big-data-analytics jupyter-notebook cognitive services industry solutions

A powerful app for processing DJI infrared (IR) images, featuring advanced thermal analysis, conversion, and visualization capabilities. Supports DJI Matrice 4 (M4T), Matrice 30 (M30T), Mavic 3 (M3T), and Mavic 2 Enterprise Advanced drones

LeManchot-Analysis is a system for abnormal detection in coupled visible-thermal images

LeManchot is a multimodal data platform including data acquisition and data transmission components

LeManchot-Points is a system for fusion of multi-modal data captured by our data acquisition system

Universal industrial anomaly detection GUI - Zero-shot defect detection for manufacturing quality control

基于PyTorch U-Net的图像边缘分割系统，支持自定义数据集训练、推理与评估，适用于工业检测、医学影像和遥感影像边缘提取

A benchmark dataset for detecting fresco print defects on cylindrical paper bobbins using Deep Learning and CNNs.

Datasets: we curatesome representative anomaly detection datasets, they are used in our current research.

Surface defect classification for industrial visual inspection using PyTorch and FastAPI.

Vision Foundation Model for industrial anomaly detection using DINOv2-ViT-Base with LoRA adaptation. Optimized through 4-version experimental study on MVTec-AD2 dataset, achieving stable performance across all 7 categories (average AUC: 0.5626).

OrbitCore: Perspective- and illumination-invariant PatchCore for dynamic environments.

Curated list of MLLM-based methods for industrial anomaly detection — taxonomy, benchmarks, research gaps

Hybrid Deep Learning Framework for Automated Spacecraft Component Defect Detection and Classification using CNN and SE Attention.

🔍 Detect anomalies in industrial settings effortlessly with a zero-shot GUI. Identify defects in real-time without training on labeled examples.

Anomaly detection pipeline for industrial surface inspection using the MVTec AD dataset. Includes classical computer vision baselines, autoencoder models, and PaDiM-based feature embeddings for detecting subtle manufacturing defects.

This repository houses the official code for RGAE+PNNR, a unified anomaly detection framework.