Course Objective & Scope
In industrial quality inspection and equipment/system fault diagnostics, defective samples are scarce and novel defect modes appear without warning. This programme covers normal-only visual and signal anomaly localisation (PatchCore, PaDiM, Reverse Distillation with Anomalib), supervised defect detection and pixel segmentation (YOLO, U-Net), and 1D-CNN/temporal autoencoder models for high-frequency vibration, acoustic and sensor waveforms.
Robustness to lighting and fixture shifts, synthetic defect generation (CutPaste/diffusion), image- and pixel-level AUROC/AUPRO metrics and line-speed edge inference are practised hands-on. Participants build an end-to-end defect detection and localisation prototype on an inspection or fault dataset.
Who Should Attend?
- Data scientists and ML engineers in manufacturing, automotive, electronics, textiles and energy
- Visual quality inspection, non-destructive testing (NDT) and automation engineers
- R&D teams diagnosing faults from sensor waveforms, acoustics and system logs
- Prerequisite: Python and foundational deep learning experience in PyTorch
Key Takeaways
- Train normal-only visual defect localisation models with Anomalib (PatchCore, PaDiM) to produce pixel heatmaps
- Build supervised object detection (YOLO) and segmentation (U-Net) pipelines for catalogued defect classes
- Detect mechanical and electrical faults from vibration and acoustic waveforms using spectrograms, 1D-CNNs and autoencoders
- Augment scarce defect classes via CutPaste and synthetic defect injection
- Calibrate pass/fail thresholds against false-scrap and escaped-defect costs at production line speeds
Curriculum
Industrial Defect and Fault Formulations
- Supervised defect classification/segmentation vs normal-only (one-class) learning
- Visual surface defects, waveform faults and log-sequence failures
Visual Defect Localisation: PatchCore, PaDiM and Anomalib
- Pretrained backbone patch features, coreset memory banks and Mahalanobis scoring
- Pixel-level anomaly heatmaps and AUPRO/AUROC evaluation
Supervised Detection, Segmentation and Synthetic Defects
- High-resolution tiling, YOLO and U-Net for tiny surface defects
- CutPaste and synthetic defect blending for rare-class augmentation
Acoustic, Vibration and Log-Sequence Fault Detection
- 1D/2D CNNs and TCNs on raw waveforms and Mel-spectrograms
- LSTM/Transformer Autoencoders and DeepLog for sequence faults
Calibration, Sensor Shift and Line-Speed Deployment
- Handling lighting/batch shifts and controlling false-reject (overkill) rates
- Millisecond inline inference with ONNX and TensorRT
Other Courses in This Category
-
Multivariate, long-horizon and probabilistic time-series forecasting with N-BEATS, N-HiTS, Temporal Fusion Transformer (TFT), PatchTST, DeepAR and TCN using NeuralForecast and PyTorch Forecasting.
- Live Course
- 5 modules
-
Large-scale deep recommendation architectures: Two-Tower retrieval with ANN (FAISS/Qdrant), sequential session recommenders (SASRec/BERT4Rec), DeepFM/DCN-v2/DLRM ranking and multi-task MMoE learning.
- Live Course
- 5 modules
-
Text preprocessing, Turkish morphology, regex, tokenisation, TF-IDF, word/sentence embeddings and building first text classification, NER and sentiment pipelines with spaCy, scikit-learn and Hugging Face.
- Live Course
- 5 modules
-
Advanced NLP under scarce labels: domain-adaptive pretraining (MLM), DeBERTa/RoBERTa fine-tuning, LoRA/PEFT, few-shot learning with SetFit, weak supervision and label cleaning (Cleanlab), long documents and cross-lingual transfer.
- Live Course
- 5 modules
-
End-to-end neural NLP covering subword tokenisation, BiLSTM-CRF and Transformer (BERT, DeBERTa, T5) architectures, with hands-on sequence tagging, seq2seq summarisation and question answering in PyTorch and Hugging Face.
- Live Course
- 5 modules
-
Running NLP models in production at low latency and cost: ONNX/Optimum quantisation, Triton/FastAPI serving, dynamic batching, PII redaction (Presidio), behavioural testing (CheckList) and text drift monitoring.
- Live Course
- 5 modules
Get in touch
