One-phase neural-network image extraction solution
A one-phase neural-network approach for image extraction that simplifies processes by integrating bounding box selection and object recognition, showcasing high accuracy with methods like fast-RNN, faster-R-CNN, RetinaNet, and Yolo.
Overview
This solution leverages a one-phase neural-network approach to streamline image extraction processes. By integrating bounding box selection and object recognition into a single phase, it addresses challenges in traditional two-phase methods. Recent advancements such as fast-RNN, faster-R-CNN, RetinaNet, and Yolo have demonstrated significant improvements in accuracy and efficiency, making this approach highly effective for tasks involving complex image and video data.
Technical specifications
Key features:
- Combines bounding box selection and object recognition in one phase for efficiency
- Utilizes advanced neural-network models like faster-R-CNN and Yolo for high accuracy
- Capable of handling partially obstructed objects and variable visibility conditions
- Incorporates techniques for synthetic data generation to enhance training datasets
- Employs hyper-parameter tuning to optimize model performance and achieve high accuracy
Technology readiness level
This image extraction technology is at TRL 6, indicating it has been demonstrated in relevant environments and is undergoing further validation through data collection, model building, and hyperparameter optimization to ensure readiness for broader applications.
About California Polytechnic State University, San Luis Obispo
California Polytechnic State University (Cal Poly) is a comprehensive public polytechnic in San Luis Obispo known for applied learning and industry-relevant research. Facilities include extensive maker and prototyping spaces, pilot-scale labs for product and process development, and an on-campus research park that co-locates companies with faculty and students. Partnerships grow through sponsored senior design, internships, and capstone collaborations, reinforced by Learn by Doing and a Central Coast location between the Bay Area and Southern California. Research is supported by competitive federal and state funding, including the National Science Foundation and the U.S. Department of Agriculture, plus corporate contracts. A dedicated technology transfer office and entrepreneurship programs guide IP, licensing, and incubation.
delivered to your inbox.