Real-Time Object Detection System

Object detection for pictures, video and webcam with a pre-trained MobileNet-SSD network: boxes, counts, alert rule, snapshots, recording, whole-video analysis and a log. Python, OpenCV, Tkinter. Works offline.

  • Price: ₹1,499
  • Category: AI / ML
  • Technology: Python, OpenCV (DNN), MobileNet-SSD, NumPy, Tkinter, Pillow, SQLite, Matplotlib
  • Includes: Full source code
  • Suitable for: BCA / B.Sc CS, MCA / M.Sc CS, B.E / B.Tech CSE & IT

Project overview

This is a desktop application for Windows with a login. The Detect screen opens a picture (13 demo photos are included) and shows the boxes, the number of objects and a list with the confidence of each. The Live screen plays a video file at its own speed or the webcam, draws the boxes frame by frame, shows the frames per second and the counts, can raise an alert when at least N objects of a chosen kind are in the picture, saves a snapshot and can record the result as a video. The Analyse screen checks every N-th frame of a whole video and draws the number of objects over time and the most seen at the same time, with a CSV export. The Reports screen lists every check and every alert and charts the objects counted. Confidence (0.1 to 0.95) and an object filter are on every screen. The network is a pre-trained MobileNet-SSD (20 object kinds from PASCAL VOC), so nothing needs training or downloading. 21 automatic code tests and 54 screen checks are included.

Modules

  1. Detect (pictures): Open a picture or a demo photo; boxes, labels and counts; list of objects with confidence; save the result picture.
  2. Live (camera or video): Frame-by-frame detection from the webcam or a video file with frames per second and live counts; alert rule; snapshot; optional recording of the result video.
  3. Analyse (whole video): Detects in every N-th frame; chart of objects over time and of the most seen objects; CSV export.
  4. Detection engine: OpenCV DNN with MobileNet-SSD; confidence filter; object filter; non-maximum suppression so one object gets one box.
  5. Reports and log: Log of every check and every alert in SQLite; chart of objects counted; CSV export; clear the log.
  6. Login: Admin login with a salted hash and a change-password dialog.

Requirements

  • Python 3.11
  • Internet once, to install the libraries (about 3 minutes)
  • Windows 10 / 11

How to run this project

  1. Install Python 3.11 from python.org and tick "Add python.exe to PATH" during setup.
  2. Extract the zip and open the project folder (STUDENT_GUIDE.md explains every step in simple words).
  3. Double-click `1_INSTALL.bat` once and wait until it says [OK] (about 3 minutes; it downloads the libraries into a private .venv folder).
  4. Double-click `2_RUN.bat` to start the project. If it does not start, double-click `CHECK_SETUP.bat` and read what it says.
  5. Log in with admin / admin123.
  6. Analyse: press 'Analyse the whole video' and explain the two charts; export the CSV. Reports: show the log of all checks and the alert list; export the CSV.

More AI / ML projects