Invention Grant
- Patent Title: Feature pyramids for object detection
-
Application No.: US17386879Application Date: 2021-07-28
-
Publication No.: US11954175B2Publication Date: 2024-04-09
- Inventor: Fangyi Chen , Chenchen Zhu , Zhiqiang Shen , Han Zhang , Marios Savvides
- Applicant: CARNEGIE MELLON UNIVERSITY
- Applicant Address: US PA Pittsburgh
- Assignee: Carnegie Mellon University
- Current Assignee: Carnegie Mellon University
- Current Assignee Address: US PA Pittsburgh
- Agency: KDW Firm PLLC
- Main IPC: G06F18/214
- IPC: G06F18/214 ; G06F18/213 ; G06F18/2431 ; G06N5/04 ; G06N20/00

Abstract:
Disclosed herein is an improvement to prior art feature pyramids for general object detection that inserts a simple norm calibration (NC) operation between the feature pyramids and detection head to alleviate and balance the norm bias caused by feature pyramid network (FPN) and which leverages an enhanced multi-feature selective strategy (MS) during training to assign the ground-truth to one or more levels of the feature pyramid.
Information query