Geometric Surface Image Prediction for Image Recognition Enhancement

This work presents a method to predict geometric surface image from photograph assist in recognition. To recognize objects, several images different conditions are required for training model or fine-tuning pre-trained model. In this work, is introduced as better representation than its color counterpart overcome lighting conditions. The predicted image. do so, the together with photographs firstly trained Generative Adversarial Networks (GAN) generator then used input evaluation on case study of an amulet recognition shows that contain less ambiguity their under and can be effectively assisting task.