Object Detection vs Object Recognition vs Image Segmentation
Object recognition:
Object recognition is the technique of recognizing the object present in images and videos. It is one of the most important applications of machine learning and deep learning. When humans look at a photograph or watch a video, we can readily spot people, objects, scenes in the same wat the goal of this field is to teach machines to understand the content of an image naturally as humans do.
Object recognition is a key skill in driverless cars, enabling them to recognize a stop sign or to differentiate a pedestrian from a lamppost. It is also advantageous in a various application such as disease identification in bioimaging, industrial inspection, and robotic vision.
Object Recognition Using Deep Learning
The most widely used method for object recognition is Convolution Neural Network (CNN). This CNN network takes an image as input and outputs the probability of the different classes. If the object presents in the image then its output probability is high else the output probability of the rest of classes is either negligible or low. The advantage of Deep learning is that we don’t need to do feature extraction from data as compared to machine learning.
Challenges of Object Recognition:
- The CNN model is a single class label and hence this approach will not work if more than one class labels are present in the image.
- We need to try a different approach If we want to localize the presence of an object in the bounding box.
Image Classification:
Image classification takes an image as an input and outputs the images classification label of with some metric (probability, loss, accuracy, etc). Example shown below.
Object Localization:
This algorithm locates object in an image and represents its presence with bounding box. Here the bounding box is in rectangular shape.
Object Detection:
This algorithm is a combination of image classification and object localization. This algorithm is proficient in dealing with multi-class classification, localization and also deals with multiple occurrences of objects in an image. It takes image as input and outputs one or more bounding boxes with the class label attached to each bounding box.
Challenges of Object Detection:
- This algorithm cannot help in determining the shape of the object if the objects cannot curve part because the bounding boxes are rectangular in shape.
- It cannot estimate few measurements such as the area of an object, perimeter of an object from image.
Image Segmentation:
Image segmentation is just an extension of object detection in which it marks the presence of an object with pixel-wise masks generated for each object in the image. This can help us in determining the shape of each object present in the image rather than bounding box approach.
Image segmentation is primarily of two types.
Semantic Segmentation:it uses different colors for labeling each pixel in the image based on the category class or class model.
Instance Segmentation:it used different colors for labeling the object pixels by identifying the boundaries of the object.
Example shown below.
I hope this article helped in clarifying Object Detection vs Object Recognition vs Image Segmentation.
