About Me
I am Wang Wei, a senior algorithm engineer with 11 years of experience, focusing on computer vision, deep learning, and embedded systems development. I am proficient in Python and C/C++ programming languages, familiar with Linux operating systems, as well as NVIDIA Jetson, Ambarella, and other embedded platforms.
Throughout my career, I have focused on solving complex computer vision problems, including object detection, object tracking, face detection and recognition, gesture recognition, and more. I am familiar with deep learning frameworks such as PyTorch and TensorFlow, and skilled at applying the latest research results to practical engineering projects.
In addition to technical abilities, I am also skilled in team management and communication, capable of effectively integrating resources and driving projects to successful completion. I am passionate about technological innovation and continuously follow the latest developments in artificial intelligence and computer vision.
Professional Skills
Programming Languages
- Python (Proficient)
- C/C++ (Proficient)
- Shell Scripting
Deep Learning
- PyTorch
- TensorFlow
- Object Detection (YOLO Series)
- Pose Estimation
- Face Recognition
- Label Studio Ml Automatic labeling
Traditional Computer Vision
- OpenCV
- Image Processing
- Feature Extraction
- Object Tracking (EKF, KF)
Embedded Systems
- Linux Systems
- NVIDIA Jetson Platform
- Ambarella Chips
- TensorRT Optimization
Tools & Platforms
- CMake, Bazel, Make
- GDB Debugging Tool
- Docker Containerization
- Git Version Control
Other Skills
- Team Management
- Project Planning
- Patent Writing
- Technical Documentation
Work Experience
Beijing Invo Automotive Technology Co., Ltd.
Visual Perception Algorithm Engineer
Responsible for DMS project development, implementing static and dynamic gesture recognition (95%+ accuracy) and facial attribute expression recognition (98%+ accuracy); deployed algorithms on Ambarella chips; developed rearview mirror contamination detection and parking spot detection systems; researched industrial defect glue detection algorithms.
Beijing Happy Eggplant Information Technology Co., Ltd.
Algorithm Engineer
Built Milvus-based distributed search engine; developed ad algorithm model_server service and MA strategy algorithm operators; implemented ad algorithm feature extraction service; trained CTR/CVR prediction models for ad algorithms.
Beijing Huixiang Technology Co., Ltd.
Image Recognition Algorithm Engineer
Responsible for image object detection and tracking algorithm training and selection; deployed algorithms on x86 and NVIDIA Jetson ARM platforms; researched 3D camera background separation and point cloud clustering algorithms; developed object tracking system based on Kalman filter and Hungarian algorithm.
Hengxin Oriental Culture Co., Ltd.
Algorithm Engineer
Researched computer vision algorithms, including traditional machine learning methods and deep learning methods; responsible for project-related paper reading and implementation, optimization of network structures; deployed algorithms on Linux and Windows platforms; developed Android mobile SDK; responsible for face beautification technology development.
Early Career Experience
C++ Engineer
Worked as a C++ engineer at Aiyiwa (Beijing) Technology Co., Ltd. and Beijing Yixiong Information Technology Co., Ltd., responsible for Linux and Android system low-level development, image processing algorithm porting, and protocol development.
Projects
Driver Monitoring System
Developed high-precision static and dynamic gesture recognition algorithms, achieving over 95% recognition accuracy. Static gestures use classification methods, while dynamic gestures use trajectory fitting and Kalman filter tracking. Also implemented facial attribute and expression recognition functions with recognition accuracy exceeding 98%.
Parking Space Detection System Based on Point Cloud and Vision Fusion
Developed a parking space detection system that fuses LiDAR point cloud data and vision segmentation images. The system extracts parking slot corner information through point cloud processing and achieves high-precision parking space localization by incorporating vehicle CAN bus parameters. The processing pipeline includes point cloud preprocessing, ground segmentation, parking line detection, and corner point extraction, ultimately outputting parking space location and type information to provide critical support for automatic parking systems.
Industrial Vision Glue Dispensing Defect Detection System
Developed a deep learning-based glue dispensing defect detection system capable of identifying various defect types including broken glue lines, excess glue, insufficient glue, and bubbles, with a detection accuracy of 97%, significantly improving production quality and efficiency. The system supports automatic threshold parameter adjustment to adapt to different production environments and condition changes.
Advertising Algorithm Platform
Built a Milvus-based distributed search engine, developed ad algorithm model_server service and MA strategy algorithms, implemented feature extraction services, and trained efficient CTR/CVR prediction models, significantly improving ad delivery effectiveness.
Face Detection and Beautification System
Developed face detection and 98 key point localization algorithms, implemented Hourglass network and GAN adversarial concepts for facial key point detection using PyTorch. Researched and implemented the facial feature point detection method from the "Look at Boundary" paper, and completed algorithm transplantation to Android mobile platforms.
Contact Me
Contact Information
w454940304@outlook.com
github: jackweiwang
Beijing