HANOI UNIVERSITY OF CIVIL ENGINEERING

Final project: Handwritten text recognition

Bac Ha Software Co., Ltd.

OCR PDF, Vision Language Model (VLM), Large Language Model (LLM)

Generative AI: Stable Diffusion (Prompt Engineering)

STE TECHNOLOGY JSC

Solve the problem of handwritten text recognition

Technology: Stable Diffusion, Prompt Engineering, Diffusers, Scheduler Algorithms

Pipeline: 1. Stable Diffusion basics -> 2. Prompt Engineering -> 3. Training a custom model -> 4. Image to image -> 5. Inpainting - exchaning classes -> 6. ControlNet

Result: Image generation according to the prompt

Technology: Pytorch, Embedding Layer, EmbeddingBag, BoW, TF-IDF

Pipeline: Text Preprocessing -> EDA -> Representation -> Modelling -> Deployment

Result: BOW: 0.8779; TF-IDF: 0.8822

(1) Programming -> (2) Calculus and Linear Algebra, and theirs Applications to AI -> (3) Probability and Statistics, and theirs Applications to AI -> (4) Machine Learning and Data Science -> (5) Steps to Deep Learning -> (6) Deep Learning for Different Data -> (7) Advanced Module Architectures -> (8) Deep Learning Applications -> (9) Generative Models -> (10) Advanced Deep Learning.

TF 07

Leader and Training

Leading and guiding a team of 11 members in the field of AI

Development orientation, guidance on participating in competitions, and writing articles about AI (Computer Vision and NLP)