Course Objective & Scope
Invoices, contracts, bank slips and scanned forms rarely follow a single template, so raw OCR text alone is seldom enough to drive business rules. Starting from image enhancement and deskewing, this programme covers modern OCR engines (PaddleOCR, Surya, Tesseract) together with visual-textual layout models (LayoutLMv3, Donut, Docling) end to end.
Multi-column pages, complex tables, signature/stamp regions and key-value extraction are practised in hands-on labs. Working on sample document types from their own organisation, participants design a Document AI pipeline complete with validation rules and human-in-the-loop review.
Who Should Attend?
- Software and AI engineers building document automation and extraction services
- Data scientists digitising finance, insurance, legal and operations workflows
- Specialists responsible for technical architecture in RPA and enterprise workflow projects
- Prerequisite: Python programming and basic deep learning knowledge
Key Takeaways
- Enhance scanned and photographed documents for noise and perspective before OCR
- Select an OCR engine for Turkish and multilingual documents and measure speed/accuracy trade-offs
- Detect page layout and table structures and convert them into JSON or database schemas
- Extract form and invoice fields using LayoutLMv3 or OCR-free Donut models
- Evaluate an end-to-end Document AI pipeline with CER/WER and field-level F1 metrics
Curriculum
Document Preprocessing and Image Enhancement
- Deskewing, denoising, adaptive thresholding and perspective correction
- PDF layer inspection and standardising image resolution
Modern OCR Architectures: Text Detection and Recognition
- Text region detection with DBNet and CRAFT, character recognition with CRNN and SVTR
- Turkish character and handwriting performance with PaddleOCR, Surya and Tesseract
Page Layout Analysis and Table Extraction
- Separating headers, paragraphs, tables, signatures and stamps via object detection
- Producing structured data from borderless and merged-cell tables (Docling, Table Transformer)
Multimodal Document Understanding: LayoutLMv3 and Donut
- Combining visual and textual signals with 2D positional embeddings
- OCR-free approaches and schema-free invoice/contract field extraction with VLMs
Production Document AI Pipelines and Quality Control
- Evaluating models with CER, WER and field-level F1
- Confidence thresholds, business-rule validation and human-in-the-loop routing
Other Courses in This Category
-
RTSP video streams, GStreamer/FFmpeg, real-time detection, multi-object tracking with ByteTrack/BoT-SORT, speed/crowd analytics and NVIDIA DeepStream, for teams extracting live operational insights from camera networks.
- Live Course
- 5 modules
-
Human action, ergonomics and anomaly detection with 3D CNNs (SlowFast, X3D), VideoMAE/TimeSformer and skeleton-based models (MMPose, ST-GCN), for teams building workplace safety, assembly-line and behaviour analysis systems.
- Live Course
- 5 modules
-
Shared embedding spaces with CLIP/SigLIP, Qwen2.5-VL, LLaVA and InternVL architectures, visual QA, multimodal RAG and QLoRA fine-tuning, for engineers building systems that reason across images and text.
- Live Course
- 5 modules
-
Surface defect detection on production lines, unsupervised anomaly detection with anomalib (PatchCore, PaDiM, EfficientAd), industrial camera/lighting design and PLC integration, for quality and manufacturing engineering teams.
- Live Course
- 5 modules
-
Inpatient fall prevention, ICU monitoring, hand-hygiene and PPE compliance, operating-room instrument counting and privacy-preserving edge anonymisation, for hospital digital transformation and healthtech teams.
- Live Course
- 5 modules
-
DICOM/NIfTI processing, 2D/3D lesion and organ segmentation on CT/MRI/X-ray with MONAI and nnU-Net, whole-slide pathology (WSI) analysis and clinical validation, for medical imaging and biomedical AI teams.
- Live Course
- 5 modules
Get in touch
