ObjectDetectionUsingTrafficVideo

Overview

This project performs real-time object detection on a traffic video using SSD MobileNet V3 with OpenCV’s DNN module. It identifies objects such as pedestrians, vehicles, and traffic signs, drawing bounding boxes with confidence scores.

Features

• Uses SSD MobileNet V3 for object detection
• Reads class labels from LABELS.txt
• Processes video frame-by-frame
• Draws bounding boxes with confidence scores
• Stops detection when 'q' is pressed

Requirements

• Python 3.x
• OpenCV (cv2)

Installation

1. Clone the repository:

   git clone https://github.com/your-repo/ObjectDetectionTraffic.git

2. Navigate to the project directory:

   cd ObjectDetectionTraffic

3. Install dependencies:

   pip install opencv-python

Usage

1. Ensure you have the following files:

   • ssd_mobilenet_v3_large_coco_2020_01_14.pbtxt (Model Configuration)
   • frozen_inference_graph.pb (Trained Model Weights)
   • LABELS.txt (Class Labels)
   • Input video file (e.g., mixkit-busy-street-in-the-city-4000-medium.mp4)

2. Run the detection script:

   python detect_video.py

Code Explanation

• Load the model:

  model = cv2.dnn.DetectionModel(frozen_model, config_file)

• Read class labels:

  with open(file_name, 'rt') as fpt:
      classLabels = fpt.read().rstrip('\n').split('\n')

• Initialize video capture:

  cap = cv2.VideoCapture('video.mp4')

• Preprocess frames for the model:

  model.setInputSize(320,320)
  model.setInputScale(1.0/127.5)
  model.setInputMean((127.5,127.5,127.5))
  model.setInputSwapRB(True)

• Perform detection on each frame:

  ClassIndex, confidence, bbox = model.detect(frame, confThreshold=0.5)

• Draw bounding boxes and labels:

  for ClassInd, conf, boxes in zip(ClassIndex.flatten(), confidence.flatten(), bbox):
      cv2.rectangle(frame, (boxes[0], boxes[1]), (boxes[0]+boxes[2], boxes[1]+boxes[3]), (255, 0, 0), 2)
      text = f"{classLabels[ClassInd-1]}: {conf:.2f}"
      cv2.putText(frame, text, (boxes[0], boxes[1]-10), font, fontScale=font_scale, color=(0,255,0), thickness=3)

• Display the video output:

  cv2.imshow('Video Object Detection', frame)

• Exit detection when 'q' is pressed:

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

Sample Output

The script detects objects in the video and overlays bounding boxes and labels in real-time.