Timeframe:
Fall 2020 – Spring 2021
Students:
Uma Kamatchi Kesava Pillai
Faculty in Collaboration:
Dr. Bill Stapleton, Dr. Harold Stern
Overview:
The goal is to develop a Deep Learning – Computer Vision approach to classify the vehicle’s colors, type to better match the vehicle’s license plate from camera feeds under different weather conditions, and correlate matches involved in these emergency alerts for the safe return of a child or elderly person.
Stages
Phase 1:
This phase is to design a Deep Learning model to classify the vehicle’s colors, types and recognize each vehicle’s license plate from camera feeds under different weather conditions, and to find possible matches involved in these emergency alerts for the safe return of a child and older adult. Vehicle types include seven classes such as SUV, Sedan, Truck, Bus, Microbus, Minivan, and Motorcycle. Vehicle colors include eight classes: green, blue, black, white, gray, yellow, white, and red. A shallow and deeper CNN, VGG16 inspired, techniques are investigated.
Phase 2:
YOLO-3 is the detection technique being investigated to bound-box the vehicles that are being classified. The detection is part of the visual aspect of the project that provides the analysis of the vehicles coming into the camera’s closes point for its final detection and classification comparison to the emergency messages.
Phase 3:
Deep learning OCR is being investigated to convert license plate information to text. The extracted text information is compared to the message itself after the color and type of the vehicle have first been cleared for further comparison to the emergency message. Application developed to test all classification and detection models when user-input values are compared to a set of test images.
Publications:
- U. K. K. Pillai and D. Valles, “Vehicle Type and Color Classification and Detection for Amber and Silver Alert Emergencies Using Machine Learning,” 2020 IEEE International IoT, Electronics, and Mechatronics Conference (IEMTRONICS), Vancouver, BC, Canada, 2020, pp. 1-5, doi: 10.1109/IEMTRONICS51293.2020.9216368.
- U. K. K. Pillai and D. Valles, “An Initial Deep CNN Design Approach for Identification of Vehicle Color and Type for Amber and Silver Alerts,” 2021 11th Annual Computing and Communication Workshop and Conference (CCWC), NV, USA, 2021, pp. 0903-0908, doi: 10.1109/CCWC51732.2021.9375917.
Poster:
- Uma K. K. Pillai, D. Valles, “Vehicle Colors and Types Detection for Amber and Silver Alert Emergencies Using Machine Learning,” Women in Science and Engineering (WiSE) Annual Conference, Texas State University, San Marcos, TX, 2020.
Thesis:
- Kesava Pillai, U. K. (2021). Vehicle types and color detection for Amber and Silver alert emergencies using machine learning (Unpublished thesis). Texas State University, San Marcos, Texas.
GitHub:
