Jewook Lee

Title: Jewook Lee: Innovator in Weakly Supervised Object Localization

Introduction

Jewook Lee, an accomplished inventor based in Seoul, South Korea, has made significant contributions to the field of computer vision through his innovative work on object localization. With a total of 2 patents to his name, Lee's inventions focus on employing weakly supervised learning methodologies to enhance object detection and localization in images.

Latest Patents

Jewook Lee's latest patents include the following:

1. **Weakly Supervised Object Localization Apparatus and Method**

This innovative apparatus is designed to improve the process of weakly supervised object localization. It comprises several key components:

- A feature map generator that creates a feature map by performing a convolution operation on the input image.

- An erased feature map generator that produces an attention map and an erased feature map through a masking operation using the attention map.

- A final map generator that generates both a final feature map and a final erased feature map through another convolution operation.

- A contrastive guidance determiner, which establishes contrastive guidance for the foreground object in the input image based on the generated feature maps.

2. **CAM-Based Weakly Supervised Object Localization Device and Method**

This patented device employs a convolutional neural network (CNN) to facilitate weakly supervised object localization. It includes:

- A feature map extractor to capture the feature map from the last convolutional layer of the CNN.

- A weight vector binarization unit that binarizes the weight vector of a linear layer during the sequential processing of the feature map.

- A feature map binarization unit that secondarily binarizes the feature map based on the previously binarized weight vector.

- A class activation map generation unit that creates a class activation map to assist in object localization.

Career Highlights

Jewook Lee currently works at the University Industry Foundation (UIF) in Yonsei University, where he engages in groundbreaking research and development in the area of intelligent systems. Throughout his career, Lee has demonstrated a commitment to pushing the boundaries of technology, particularly in the domain of artificial intelligence and machine learning.

Collaborations

In his research endeavors, Jewook collaborates with talented individuals such as Hye Ran Byun and Sanghuk Lee. These partnerships foster an environment of innovation, allowing for the exchange of ideas and expertise that advances their shared goals in the realm of object localization and computer vision.

Conclusion

Jewook Lee exemplifies the spirit of innovation through his contributions to weakly supervised object localization. With his latest patents and collaborative efforts at the University Industry Foundation, he continues to make strides that reshape the landscape of computer vision technology. His work not only pushes the envelope of current methodologies but also opens doors for future advancements in the field.