Description:TECHNICAL FIELD:
[0001] This disclosure belongs to the field of generation of image datasets and in particular to the field of image dataset generation for use in a system for computer vision applications.
BACKGROUND:
[0002] In the field of computer science and in particular, in artificial intelligence and supervised machine learning, various types of datasets are required to train models. Such datasets are called training datasets. Once the model is trained, the model may be ready to operate on data other than from the training data set, that is, data from the real world. The result of such an operation by the trained model may be diagnosis, prediction, classification and so on depending on what the model has been trained to do. Generally, the larger the dataset on which a model is trained, better is it ready to operate on real-life data. It is possible that once the model starts operating on real-life data, the model continues to learn and refine itself as the real-life data on which it operates acts as an , Claims:WE CLAIM:

1. A method for generating a plurality of images forming an image dataset for use in computer vision applications, the method comprising:
receiving, by an input module (105), one or more seed images and one or more input texts defining the image dataset, wherein the one or more input texts comprise at least one or more of a text defining a feature of the one or more seed images to be retained, a text defining a feature of the one or more seed images to be modified, a text defining a feature to be introduced into the one or more seed images, and a text defining at least one feature of each of the generated plurality of images that is to be segmented;
generating, by an image generation module (110) using a first pre-trained generative adversarial network (GAN) based module, a plurality of images having the features of the one or more seed images based on the one or more seed images and the one or more input texts, wherein the features of the one or more seed images includes one or more of a feature