Abstract: Detecting fine-grained similarity in image includes determining a core area of a search image by generating an image salient map from a plurality of layers of the search image and determining a connected area based on the image salient map. Feature descriptors are generated from the core area of the search image. A plurality of capsule vectors are generated from different ones of a plurality of keypoints of the feature descriptors. Capsule vectors of the search image are compared with capsule vectors of each image of the dataset to generate a top-K matrix. Similarity scores for the top-K matrix are calculated. One or more image of the dataset having fine-grained similarity with the search image are selected based a bundled similarity score for each image of the dataset. The bundled similarity score is a summation of the similarity scores of the image.

Abstract: One or more systems, devices, computer program products and/or computer-implemented methods of use provided herein relate to detecting a closed ring in a three-dimensional (3D) point cloud via cycle basis. A system can comprise a memory configured to store computer executable components; and a processor configured to execute the computer executable components stored in the memory, wherein the computer executable components can comprise a filtering component that can filter a first undirected graph of a three-dimensional (3D) point cloud, by eliminating one or more edges of the first undirected graph that are longer than an adaptive threshold, wherein filtering the first undirected graph can produce a second undirected graph; and a detection component that can detect a minimum cycle basis of the second undirected graph to determine a cycle path that can traverse an irregular annular shape that is represented by the 3D point cloud.

Abstract: A computer-implemented method, system and computer program product for improving accuracy of a vision model. Images of an object with a first set of perspectives are received from a dataset used to train the vision model. A three-dimensional model of the object is then generated using the images of the object from the dataset. Using the three-dimensional model of the object, images of the object with a second set of perspectives are obtained. For example, the second set of perspectives may include different perspectives than the perspectives of the object from the images contained in the dataset. The dataset used to train the vision model may then be augmented with such images of the object with a second set of perspectives. In this manner, the dataset used to train the vision model includes a greater number of perspectives of the object thereby improving the accuracy of the vision model.

Abstract: In an approach for detecting vehicle identity and analyzing damage status using a single video, a processor provides an instruction for taking a video of a vehicle for a damage evaluation. A processor receives the video of the vehicle for the damage evaluation. A processor verifies the vehicle in the video being the same vehicle for the damage evaluation. A processor evaluates a damage status of the vehicle. A processor outputs a damage report based on the damage status of the vehicle.

Abstract: Data augmentation is described to train an artificial intelligence model that includes analyzing a first data set to measure an amount of data in the data set and the variation in the amount of data in the first data set to determine deficiencies for training an artificial intelligence model. Augmenting data is added for the first data set having an amount of data measured that fails to meet a threshold value. Deficiencies in the variation in the amount of data in the first data set are augmented using augmentation methods outside the variation scope of the first data set to provide a second data set of augmented data. An artificial intelligence model is trained with a combined data set of the first data set, and the second data set of augmented data when the first and second data set have an amount of data meeting the threshold value.

Abstract: A system and method for adaptive color assignments to image labels during annotation of datasets includes preparing a dataset for image labeling by an annotator by: leveraging a global color analyzer to perform a global color distribution of a plurality of images to identify one or more overall colors present in the plurality of images, and a local color analyzer to perform a local color distribution for each image to identify one or more colors present in an area of interest of the image, and selecting a plurality of candidate colors to be used as image labels by the annotator, based on an output of the global color analyzer and an output of the local color analyzer.

Abstract: A method, a computing system and a computer program product for collecting and labelling images includes capturing a video of an object with a camera. A movement trace of a pointer is recorded that outlines the object while capturing the video of the object. Further included is generating a labeled image based at least on the captured video of the object and the recorded movement trace of the pointer. The labeled image includes the object and a line that surrounds the object.

Abstract: In an approach for detecting vehicle identity and analyzing damage status using a single video, a processor provides an instruction for taking a video of a vehicle for a damage evaluation. A processor receives the video of the vehicle for the damage evaluation. A processor verifies the vehicle in the video being the same vehicle for the damage evaluation. A processor evaluates a damage status of the vehicle. A processor outputs a damage report based on the damage status of the vehicle.

Abstract: A method of determining a smoothness of a surface of an object is provided. The method includes processing an image of the object to obtain opposed first and second boundary lines of a first area of the image where light from a light source is reflected off the surface of the object. The method also includes resizing the image to normalize an average distance between the first and second boundary lines. The method includes processing the resized image to obtain resized first and second boundary lines. The method also includes calculating a variance from linearity for at least one of the resized first and second boundary lines to determine a smoothness value of the surface of the object.