DƯƠNG CÔNG TÍNH
Ho Chi Minh, Viet Nam
Phone: 036*******
Email:**************@*****.***
PROFESSIONAL SUMMARY
Have experience with Python, C++, Terraform, AWS, Deepstream.
Have experience with Artificial Intelligence, Machine Learning, Deep Learning, Data processing and analysis
Have experience with Data Structures and Algorithms by participating in Competitive Programming. TECHNICAL EXPERTISE
Machine Learning Linear Regression, Logistic Regression, Decision tree, KNN, SVM, Random Forest Deep Learning CNN (VGG, Resnet, Inception Net, DenseNet). LSTM, GRU,Unet Framework Tensorflow, Pytorch
Library Pandas, Numpy, Matplotlib, OpenCV
Operating Systems Windows, Linux
Orthers Git, AWS, Deepstream.
PROFESSIONAL EXPERIENCE
6/2022 – 10/2022 Company:Greystone Data Systems
Projects: Camera de-kit
Position: Machine learning Engineer
Responsibility:
- Research model segmentation to recognize and crop image and damaged position detection
- Create Data box and bag for model.
- Processing, Analysis data for model yolov5 and ResUnetPlusPlus
- Recognize box and bag:
+ Pipeline: Image ->Yolov5-> Resnet50 -> Output.
- Crop box and bag:
+ Pipeline: Image -> Yolov5 -> ResUnetPlusPlus -> OpenCV to crop and rotation image.
- Damaged position detection:
+ Pipeline crop box -> Yolov5.
- Train, Test, Improve model.
- Build Dockerfile environment to integrate model into software. Projects: Detect and Action recognition: (Research + Test) Position: Artificial Intelligence
Engineer Responsibility:
- Research patterns to Detect and compare models: Yolov5, Yolov4, SSD, Mask RCNN
- Research patterns to Track and compare models: Strongsort, Deepsort
- Build model Classification.: Transfer learning resnet50+GRU
- Choosing the right technology for Xavier NX of Nvidia
- Collect 5 category data for yolov5 and 2 class for Classification from source RTSP.
- Processing, Analysis, Selection of Data and Label for model yolov5. Analysis 2 class of model Classification, process data into 5D form for input (B, T, H, W, C) and output are 1 and 0 for 2 class
+ Pipeline Source RTSP -> Yolov5-> Strongsort with Osnet -> Classification -> Output
- Use yolov5 to detect, strongsort to Tracking and Classification to action recognition
- Deploy project on Xavier NX with SDK Deepstream of nvidia
+ Pipeline on Deepstream: 2 Source -> Stream Mux ->Primary Detector (yolov5.engine)-> Object Tracker (Deepsort + Reid) -> Secondary classifies (Classification) -> OSD -> Sink Techniques:
- Deep Learning, Pytorch, Tensorflow, Linux, Github and Deepstream 6/2021 – 4/2021 Company: Gtopia (6 months internship) Projects: Smoke and Fire detection
Position: Machine Learning Engineer Responsibility:
- Collect data from source: youtube google…
- Data processing, analysis and evaluation
- Analyze and evaluate model, improve model in accuracy and FPS.
- Researching and building models for yolov5 to detect. Resnet50 for classification Techniques:
- Deep Learning (Object detection, Classification), Pytorch, Tensorflow, Linux and Github. AWS HONOURS AND AWARDS
4/2018 Programming competition in University of Education HCMC 2018 (UP - OLP)
- My team was placed 2nd among 25 teams.