Heng Hao

Title: Innovations by Heng Hao in Machine Vision and Object Discovery

Introduction

Heng Hao is an accomplished inventor based in San Jose, CA, with a focus on advancements in machine vision and object discovery technologies. He holds a total of 3 patents that showcase his innovative contributions to the field.

Latest Patents

One of his latest patents is titled "Bayesian semantic segmentation active learning with beta approximation." This invention involves training a machine vision model using an acquisition function that identifies unlabelled pixels with high uncertainty in their predicted label values. The iterative training process requires very few labelled pixels to accurately label areas of a data image.

Another significant patent is "Object discovery using an autoencoder." This invention addresses the challenges of supervised learning by utilizing patches to discover objects in unlabeled training images. The AI-driven process replaces manual training steps with a machine-centric approach, effectively identifying frequent objects such as human faces, bodies, animals, and vehicles from unorganized data images.

Career Highlights

Heng Hao is currently employed at Samsung SDS America, Inc., where he continues to develop innovative solutions in machine vision and artificial intelligence. His work has significantly contributed to the advancement of technologies that enhance object detection and image representation.

Collaborations

Heng collaborates with talented coworkers, including Sima Didari and Jae Oh Woo, who contribute to his projects and innovations.

Conclusion

Heng Hao's contributions to machine vision and object discovery through his patents reflect his commitment to innovation in technology. His work continues to influence the field and pave the way for future advancements.