Imaging, Multimedia and Graphics (IMG Lab)

Researchers in the IMG lab work in the field of image processing and computer vision. A staggering amount of data is in visual form from a diverse range of disciplines such as medicine, agriculture, manufacturing, and archaeology.  Analysis, annotation and summarization of this data by highly-trained human experts is extremely time-consuming. The mission of the IMG lab is to develop techniques in computer vision that will aid in the automated processing of this rich visual data.

Notable Research Accomplishments

•  Automated segmentation and analysis of plant breeding trials at the microplot level with applications to plant phenotyping
• 2D and 3D plant root segmentation, inpainting, and analysis pipelines using deep neural networks and graph neural networks to characterize root system architecture for phenotyping

Current Research Projects

• Investigation of the relationship between human factors in image annotation and deep neural network model performance
• - Investigation of the effect of traditional and domain-specific semi-automatic image annotation tools on inter-annotator agreement and deep neural network model performance
• Gaussian splatting for 3D video-game art asset generation from 2D concept art
• Various projects in the area of deep learning for computational agriculture

Faculty

• Dr. Mark Eramian