Introduction to OpenLORIS-Object

OpenLORIS-Object is a dataset specifically designed for lifelong robotic vision tasks. It includes RGB-D images of 69 object instances across 19 categories, collected under various environmental conditions. A more detailed description about how the data was collected can be found in the associated paper. This dataset is particularly useful for testing the robustness of object recognition algorithms under changing conditions, which aligns well with the capabilities of Dynamic Mode Decomposition (DMD).

Project Objectives

1. Apply DMD to analyze object recognition under varying environmental conditions
2. Utilize DMD for temporal analysis of video sequences
3. Evaluate the robustness of DMD-based object recognition

Approach

• Data Loading and Preprocessing
• DMD Analysis across Environmental Factors
• Comparative Analysis Across Environmental Factors
• Reconstruction Quality Analysis