Given an image, one may want to identify an embedded image within the given image. This requires the use of an isomorphic transformation of the projective space from one image to the other. In this research, OpenCV, ORB Descriptors, FLANN, Homography/Affine Transformation along with template matching methods were used to do direct image matching via feature and keypoint matching for scale-variant images. ORB was utilized instead of SIFT to keep efficiency and reduce cost. This also included the research of a multi-layer convolutional architecture via DeepMatching