Computer Vision Machine Learning

Vu Quoc Hung Email : adz90o@r.postjobfree.com

Linkedin: https://www.linkedin.com/in/gnuhcouq/ Mobile : 038*-***-*** Github: https://github.com/hung1012 Date of birth: 10/12/2001 Education

•

Hanoi University of Science and Technology 1 Dai Co Viet, Hai Ba Trung, Hanoi Major: Computer Science; CPA: 3.5 Aug 2019 - Aug 2023 Skills Summary

• Programming Languages: Python, C/C++, Java, Javascript, SQL

• Tools: Pytorch framework, Pandas, Tensorboard, WandB, Huggingface, TensorFlow framework, Nodejs

• Foundation knowledge: Computer Vision, Generative Model, Probabilistic, Optimization, Machine Learning Experience

•

HUST: BKAI Research Center 1 Dai Co Viet, Hai Ba Trung, Hanoi Research Assistant: Computer Vision Lab JuLy 2023 - Now

Diffusion Model for Semantic Segmentation - Segmentation Model: Utilize diffusion based model for medical semantic segmentation tasks, especially for polyps and skin cancer segmentation. Survey and experiments with models such as: SegDiff, MedSegDiff, DiffUnet. Improve model speed by converting image to latent space using VAE. Gain a 9.3 graduation thesis score in this topic.

•

Pixta Inc: Pixta Vietnam 1 Phung Chi Kien, Nghia Do, Cau Giay, Hanoi Data Scienctist - AI Research Team Sep 2022 - July 2023

Diffusion Model - Image Generative Model: Read papers about the generative model, especially Diffusion based models such as: Imagen, Dall-E2, GLIDE, Stable Diffusion, etc. Read papers and research on the VAE model, and CLIP model. Training new Stable Tag-to-Image Diffusion Model with private data and research about optimizing training and sampling speed. Research on improving model performance by substitute component: VAE, U-net, Language Module. Run inference, visualizing and comparing results of each model.

Training project: Read about Object Detection Models such as Retinanet, YOLO-based model, MobileNets, and Gradcam Visualize for pretrained model, and learn about Clean Code. Seminar weekly about the latest technology in AI: Distributed training, Optimization tools, SoTA papers, etc.

•

CMC Corporation: CMC-CIST 11 Duy Tan, Dich Vong Hau, Tu Liem, Hanoi Computer Vision Internship - AI Lab April 2022 - July 2022

OCR - Optical Character Recognition: Extracts and reconstructs text format data from scanned documents, camera images, PDFs, and other image-only documents using Deep Learning techniques and architectures. Experiment with some type of model: CRNN (Convolutional Recurrent Neural Network, using CTC loss), Attention-based OCR(Using CNN + RNN with Attention), Transformer OCR (CNN + Transformer), etc. Academic Projects

• Image Captioning: Flickr8k dataset include 8091 images with 5 text captions for each image. Model built with CNN for encoder and LSTM for decoder. Training with 30 epochs and gaining a model that generates a caption for each image. Achieve 31.1 in BLEU-1 score.

• Lazada Predict Star In Reviews: Build a model that can predict the star based on its following comment, the start value is 1,2,3,4,5. Data crawl from e-commerce platforms: Tiki, Lazada, and Shoppe, including text comments and the start given. Model using pre-trained PhoBERT with simple Softmax layer achieve 93% accuracy.

• Facial Emotional Recognition Project: FER Dataset with 28702 images and 7 classes. The model is built based on VGG-16 architecture. The trained model achieves 68% accuracy in validation

• Traffic Sign Recognition Project: Traffic-signs dataset with 34800 images and 43 classes. Model built with simple CNN architecture and Dropout to avoid over-fitting. Using data augmentation techniques to generate more images for training. Gain 89% validation accuracy for predicting traffic-sign image’s label.

• Rasa Chatbot Covid Project: Chatbot answering about Covid-19. Built based on the Rasa framework. Activities

•

Google Developer Student Clubs: GDSC - HUST Hanoi University of Science and Technology Member of Innovation team and Developer team Feb 2020 - Current

Organize monthly events, workshop with GDG Hanoi, GDG VietNam Webinar

Seminar weekly about AI, Deep Learning, Web

Training each other about Web, App, and Deep Learning and building websites and applications. Honors and Awards

• Bronze Medal for National Math Competition on Internet in 2017 Certifications

• Deep Learning Specialization Certificate - Coursera

• TensorFlow Developer Professional Certificate - Coursera

• TOEIC Certificate with score 655

Interests

• Read articles about Computer Science or Artificial Intelligence, especially articles in the field of Machine Learning, Deep Learning and Computer Vision.

Target

• Studying and researching in the field of AI and Computer Vision. Thereby applying that knowledge to build tools and applications capable of solving real-world problems.

Contact this candidate