Saeed Izadi
Papers
2
Total Citations
65
H-Index
2
About
Saeed Izadi is a computer vision researcher whose work centers on understanding how deep learning models achieve robustness to image variations. His primary research areas include object recognition, controlled dataset design, and the systematic evaluation of convolutional neural networks (CNNs). Izadi’s major contribution lies in creating and analyzing large-scale, controlled object datasets to isolate and measure specific visual factors—such as translation, scale, pose, illumination, and background—that affect model performance. His most cited work, "iLab-20M: A Large-Scale Controlled Object Dataset to Investigate Deep Learning" (2016, 63 citations), provides a unique benchmark that goes beyond traditional natural-image datasets by offering precise control over nuisance variables. This allows researchers to probe exactly how CNNs handle each type of variation. In a related study (2015), he explored what these controlled experiments reveal about the inner workings of deep networks. By bridging the gap between big data and rigorous experimental control, Izadi’s work offers foundational insights for building more robust and generalizable vision systems, making it essential reading for students and researchers interested in the limits and capabilities of deep learning.
Research Focus
Key Achievements
Top Papers
- 1
- 2