IP Library › Granted Patent US 12,227,212
Granted Patent B2
US 12,227,212 · App. 17/330,591 · Granted Feb 18, 2025

Computer vision based real-time pixel-level railroad track components detection system

Inventors: Yu Qian (Irmo, SC); Feng Guo (Columbia, SC)
Assignee: University of South Carolina
B61L23/042G06N3/08G06T7/0008G06T7/75G06T2207/20081G06T2207/20084G06T2207/20132
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,227,212
App. No.
17/330,591
Filed
May 26, 2021
Granted
Feb 18, 2025
Kind
B2
Art Unit
2668
USPC
382/108
Abstract

Systems, methods and devices for a computer vision-based pixel-level rail components detection system using an improved one-stage instance segmentation model and prior knowledge, aiming to inspect railway components in a rapid, accurate, and convenient fashion.

Claims (18)

1. An improved one-stage object detection model comprising:

at least one camera;

at least one graphics processing unit;

at least one one-stage object detection model YOLOv4-hybrid, comprising

Swish;

Leaky-ReLU; and

a combination of Mish and Swish; and

the one-stage object detection model YOLOv4-hybrid employs a hybrid activation function, which includes parameters:

precision;

recall;

mAP; and

F1 score functionality.

2. The improved one-stage object detection model of claim 1 , wherein the model functions in diverse light conditions.

3. The improved one-state object detection model of claim 1 , wherein the model can detect image modification.

4. The improved one-state object detection model of claim 3 , wherein detection of image modification includes analysis of a structural integrity of a component analyzed by the model.

5. The improved one-state object detection model of claim 1 , further comprising a deep learning algorithm.

6. The improved one-state object detection model of claim 1 , wherein the model performs real time component detection.

7. The improved one-state object detection model of claim 1 , wherein the model performs real time component detection on a railway.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded May 26, 2021
From: QIAN, YU; GUO, FENG
To: UNIVERSITY OF SOUTH CAROLINA
Reel/Frame 056355/0207 →
Continuity (2)
Provisional Application 63030376 · May 27, 2020
Related Publication 20210370993A1 · Dec 2, 2021
References Cited (7)
US 20170106885A1 · Singh · 2017 [cited by examiner]
US 20180018524A1 · Yao · 2018 [cited by examiner]
CN 112347943A · 2021 [cited by examiner]
Li, Y., Li, S., Du, H., Chen, L., Zhang, D., & Li, Y. (2020). YOLO-ACN: Focusing on Small Target and Occluded Object Detection. IEEE Access, 8, 227288-227303. (Year: 2020). [cited by examiner]
Y.-W. Lin, C.-C. Hsieh, W.-H. Huang, S.-L. Hsieh and W.-H. Hung, “Railway Track Fasteners Fault Detection using Deep Learning,” 2019 IEEE Eurasia Conference on IOT, Communication and Engineering (ECICE), Yunlin, Taiwan,… [cited by examiner]
Bochkovskiy, A., Wang, C., & Liao, H.M. (2020). YOLOv4: Optimal Speed and Accuracy of Object Detection. ArXiv, abs/2004.10934. (Year: 2020). [cited by examiner]
Li, Y., Trinh, H., Haas, N., Otto, C., & Pankanti, S. (2014). Rail Component Detection, Optimization, and Assessment for Automatic Rail Track Inspection. IEEE Transactions on Intelligent Transportation Systems, 15, 760-… [cited by examiner]