Object Detection and Extraction

Web app: https://object-detection-nd-extraction.herokuapp.com/

You can also clone the repositry and then run

git clone https://github.com/amritansh22/CaMicrosopeChallenge.git

Head over to the folder

cd CaMicrosopeChallenge

Run the app by the command

python app.py

Open a web browser and paste the following address in the search bar

http://127.0.0.5:6302/

Now upload the image on which the object detection is to be performed.

Now enter the class of the objects that are to be dedcted and press submit(All possible classes are given at the bottom of the page).

Now an image is shown with bounding boxes having objects of given class. The total number of objects in the image is given along with the total number of objects of specified class. And then the bounding box dimensions are given.

Working demo

When you head over the link above(http://127.0.0.5:6302/) you should see a page like this:

We now upload an image, having two dogs.

Select class of dog and press the button.

The output webpage look like this.

And the extracted objects are shown as

You can see that all the extracted objects are shown as diffrent images.

Had we selected a any other class we would see nothing. Say we choose class person. The ouput page looks like:

You could see there are no bounding boxes ouput here as there are no persons in the image.

Working example 2

This is the image on which we will perform object detection. We will choose the class of person in the image.

We will get an output page like this

and the bounding box dimensions like this

The images of the extracted objects looks like this