Processing Big Data: Air Quality and Weather Integration for Environmental Insights (2023)
Big Data Processing
The project focuses on integrating extensive air quality and weather datasets to perform in-depth analysis using Apache Hadoop, Apache Spark, and Apache Hive. By combining these datasets, the goal is to explore meaningful relationships between weather patterns and air quality. The implementation leverages Hadoop's distributed storage, Spark's fast data processing capabilities, and Hive's query system to handle large-scale data efficiently, enabling actionable insights and a deeper understanding of environmental dynamics.
Basic Info:
- Team: Haoyu Yang, Zeyi Song
Deep Learning
The project aims to perform human action recognition on images and evaluate the accuracy and performance of the implemented methods. Using YOLOv5 for object detection and a fine-tuned ResNet50 for feature extraction and classification, the system identifies and labels human actions captured in static images. This approach enables the recognition of actions such as running, jumping, or sitting. The project not only focuses on achieving accurate recognition but also addresses challenges like pose variations, occlusions, and diverse environments to enhance the model's robustness and reliability.
Basic Info:
- Team: Haoyu Yang, Ruijie Sun
Deep Learning: Fingertip Position Prediction Using VGG16 (2023)
Deep Learning
This project utilizes a CNN model based on VGG16 with a customized fusion layer to predict the positions of fingertip tips from RGB-D images of a robotic hand captured by three camera views. Extensive data preprocessing and transformations were conducted to ensure the robustness and accuracy of the model. By effectively integrating multi-view spatial information, the system achieves precise and reliable fingertip position predictions, demonstrating its potential for advanced robotic hand applications.