OBJECT DETECTION using Deep Neural Network

In the last few years deep learning has made an greater impact in object detection. Considering the development of technology and the growth we can able to find almost any kind of object in image or videos.

So, what is Object Detection ?

Object Detection is a technique used to identify and locate objects in photos and videos.

For Example,

Convolutional Neural Network:

We can use variety of techniques to perform Object Detection. some of the popular Deep Learning methods are Convolutional Neural Network

CNN was Inspired From Brain’s Architecture.

What is CNN ?

CNN is a class of Deep Neural networks commonly applied to examine visual imagery

Some of the most prominent Object Detection Algorithms and Methods are..

1. R-CNN,

2. Fast R-CNN

3. Faster R-CNN

RCNN:

It is a region based convolution neural network. It extracts many regions from the input image and labelling the classes and bounds it with box.

Step 1: Input an image

Step 2: Extract region proposals using algorithm like Selective Search

 Step 3: Use transfer learning, specifically feature extraction, to compute features for each proposal using the pre-trained CNN.

Step 4: Classify each proposal with SVM

Fast RCNN:

1.   In contrast to the R-CNN, the rapid R-CNN feeds the CNN the complete picture for feature extraction rather than just particular area suggestions. Additionally, this CNN may be trained

2.   These region recommendations, which come in a variety of forms, identify areas of interest on the CNN output. Then, in order to be readily concatenated, these areas of interest further extract features with the same form.

3.   The region of interest (RoI) pooling layer is something that the rapid R-CNN brings to the table.

4.   The layer receives the CNN output and the region suggestions, producing concatenated features of form nch2w2 that are then extracted for each of the area suggestions.

5.   By defining the sizes of the pooling window, padding, and stride, we may indirectly influence the output form in the pooling layer. In contrast, in the area of interest pooling layer, we may directly set the output form.

Conclusion:

Object Detection is used in wide range of industries for many purposes. It starts from smartphone devices to surveillance drones.

Happy Learning.