A framework for image classification

Image annotation process requires time and human intervention. In this research we propose a framework to incrementally annotate images in the database based on user feedback. At the beginning users provide some annotations for images manually as a ground truth. Classifier is trained based on this ground truth. The classifier predicts annotation for new images that are not part of the ground truth. Feedback is collected from the users to increase the size of the training set and then the classifier is retrained. The system strives to capture feedback from users and retrains the classifier on the new training set. Our proposed framework facilitates semi-automatic image annotation