Image Segmentation and Object Recognition
Keywords:Image Segmentation, Object Recognition, Computer Vision
The work focuses on object recognition to perform a set of closely related tasks in the field of computer vision, involving object detection and identification. Image classification involves an array of tasks like identifying the class of an object in an image. An object that is identified can also be precisely localized and a bounding box is drawn around it. Object detection combines both localization and identification and classifies one or more objects in image. For Object recognition, a technique called ‘Mask Region-based Convolution Neural Networks’ or Mask R-CNN is used. It is an extremely efficient approach for object localization and recognition tasks. It is an extension over Faster RCNN method by adding a parallel process for the prediction of a highly accurate object segmentation mask that conforms to the bounds of each detected object in the Region of Interest (RoI) by performing pixel by pixel analysis and classification while still staying highly performant. By using the above methods, individual objects can be identified more precisely than with precise localization. It is also easy to generalize the algorithm to other tasks like estimating the pose of a human in the image or applying a color filter to the image selectively on any object
How to Cite
Copyright (c) 2021 Journal of Research Proceedings
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.