Phearith/Traffic_Sign_Detection_Using_YOLOv8

Traffic Sign Detection Model

This repository contains a deep learning model for traffic sign detection. The model is trained to detect and classify traffic signs in real-time, suitable for applications like autonomous driving and advanced driver assistance systems (ADAS).

---

It suitable for :

• Computer Vision Learning
• Academic and research projects
• Autonomous driving prototypes

---

Model Details

• Model type: YOLO / PyTorch
• Task: Object detection (traffic signs)
• Dataset: German Traffic Sign Recognition Benchmarks
• Input: Images (RGB)
• Output: Bounding boxes + class labels
• Framework: PyTorch
• License: MIT

---

Dataset

The model was train on German Traffic Sign Recognition Benchmarks (GTSRB) that contain 43 class include with all the images and the all the label. And we filter some sign on the GTSRB and the overall of the filtering contain 33 class. There are 'TurnRightAhead', 'NoPassingTrucks', 'DangerousCurveLeft', 'Yield', 'SpeedLimit60' 'EndNoPassingByTrucks', 'EndNoPassing', 'RoundaboutMandatory', 'KeepLeft', 'NoPassing','KeepRight', 'RightOfWayCrossing', 'PriorityRoad', 'GoStraightOrLeft', 'Stop', 'NoVehicles', 'VehiclesOver3.5TonsProhibited', 'NoEntry', 'GeneralCaution','GoStraightOrRight', 'DangerousCurveRight', 'DoubleCurve', 'BumpyRoad', 'SlipperyRoad', 'RoadNarrowRight', 'RoadWork', 'TrafficSignals', 'Pedestrians', 'ChildrenCrossing','BicyclesCrossing', 'AheadOnly', 'WildAnimals', 'TurnLeftAhead'

---

Repository Structure

traffic_sign_model/ ├── model.pt # Model training ├── config.yaml # Model configuration ├── README.md # Documentation

---

Install dependencies

python -m venv venv
venv\Scripts\activate
source venv/bin/activate
pip install ultralytics
pip install opencv-python
pip install huggingface_hub

---

Loading the model

model = YOLO("yolov8m.pt")
results = model.train(
    data="/kaggle/input/traffic-sign-detection/dataset/data.yaml",
    epochs=30,
    imgsz=1280,
    batch=10,
    mixup=0.1,
    copy_paste=0.15,
    amp=True,
    workers=2,
    patience=15,
    name="yolov8m_traffic_1280"
)

Image Detection

import cv2
from ultralytics import YOLO

model = YOLO("./traffic_sign_model/best_yolov8m.pt")

image = cv2.imread("./image.png")

results = model(image)

annotated_frame = results[0].plot()

cv2.imshow("YOLOv8 Detection Results", annotated_frame)
cv2.waitKey(0) 
cv2.destroyAllWindows() 

Real-Time Camera Detection

from ultralytics import YOLO
import cv2

model = YOLO("./traffic_sign_model/best_yolov8m.pt")
cap = cv2.VideoCapture(0)
if not cap.isOpened():
    print("Cannot open the webcame")
    exit()

while True:
    ret, frame = cap.read()
    if not ret:
        break

    results = model.predict(source=frame, conf=0.6)
    annotated_frame = results[0].plot()

    cv2.imshow("YOLO webcame test", annotated_frame)

    if cv2.waitKey(1) & 0xFF == ord('q'):
        break
cap.release()
cv2.destroyAllWindows()

---

References

[1] Ultralytics, "YOLO Documentation," [Online]. Available: https://docs.ultralytics.com/. [Accessed: Dec. 16, 2025].

[2] R. Kumar, A. Gupta, and D. Rajeswari, "Traffic Sign Detection Using YOLOv8," TIUTIC Journal, vol. 7, Art. no. 10, 2019. [Online]. Available: https://tiutic.org/pdf/volume/Vol_7/Vol7_Article_10.pdf

[3] M. Serna and A. Ruichek, "Traffic Signs Classification by Deep Learning for Advanced Driving Assistance Systems," 2019. [Online]. Available: https://www.researchgate.net/publication/335006038

[4] Y. Wu, Y. Tian, and J. Liu, "Traffic Sign Detection Based on Convolutional Neural Networks," 2013. [Online]. Available: https://xlhu.cn/papers/Wu13.pdf

[5] A. Bochkovskiy, C. Y. Wang, and H. Y. M. Liao, "The Mapillary Traffic Sign Dataset for Detection and Classification on a Global Scale," in Proc. European Conf. on Computer Vision (ECCV), 2020. [Online]. Available: https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123680069.pdf

[6] Kaggle, "German Traffic Sign Recognition Benchmark (GTSRB) Dataset," 2019. [Online]. Available: https://www.kaggle.com/datasets/meowmeowmeowmeowmeow/gtsrb-german-traffic