Object Detection using TensorFlow Object Detection API

By Surendhar R

Object Detection is a widely used Computer Vision technique that is used to detect and predict multiple objects in images and videos in real-time.

Object Detection using TensorFlow Object Detection API

Introduction

Object Detection is one of the widely used applications of Computer Vision. Object Detection is used to detect multiple objects in the images and videos and tell what the object is. In Object Detection, multiple objects are detected using bounding boxes. We can also use Object Detection in real-time live videos and images. It has a wide range of applications in the industry such as Safety PPE kit detection in manufacturing industries, Mask Detection in Ola and Uber apps, Self-Driving Cars, Crowd Counting in public places to maintain social distance, etc.

Tech Stack 

Programming Language: Python

API: TensorFlow Object Detection API

1. TensorFlow Object Detection API is a framework created by TensorFlow to do real-time object detection tasks easily and efficiently.

2. The TensorFlow Object Detection API is built on top of the TensorFlow Library.

3. There are many pre-trained neural network models available in TensorFlow Model Zoo.

4. They are different from each other by their neural network architecture, accuracy, and execution speed of the model.

5. These available models are already trained on the different datasets such as COCO, KITTI, and Open Images Dataset.

6. We can use these pre-trained models to do object detection tasks easily and efficiently using this TensorFlow Object Detection API.

Official TensorFlow Object Detection API Installation Steps and Guide Link:

https://tensorflow-object-detection-api-tutorial.readthedocs.io/en/latest/install.html

Python Modules:

1. TensorFlow 2.0
2. OpenCV
3. OS Module
4. Object Detection Module

IDE Used

Jupyter Notebook in Anaconda Navigator

Functionality

1. Upload the Image or Video
2. It loads and pre-processes the image or video.
3. SSD Mobile Net Neural Network model apply to the image or video
4. Finally, it detects the objects in the image and the video, draws the bounding boxes, and tells what the object is.

Output:

1. 

2.