Abstract: Computer vision systems and methods for determining roof shapes from imagery using segmentation networks are provided. The system obtains an image of a structure from an image database, and determines a flat roof structure ratio and a sloped roof structure ratio of the roof structure using a neural network. Based on segmentation processing by the neural network, the system determines a flat roof structure ratio and a sloped roof structure ratio based on a portion of the roof structure classified as being flat and a portion of the roof structure classified as being sloped. Then, the system determines a ratio of each shape type of the roof structure using a neural network. The system generates a roof structure shape report indicative of a predominant shape of the roof structure and ratios of each shape type of the roof structure.

Abstract: Provided are: an image diagnosis assistance apparatus capable of assisting diagnosis of an endoscopic image captured by an endoscopist; a data collection method; an image diagnosis assistance method; and an image diagnosis assistance program. The image diagnosis assistance apparatus is provided with: a lesion assessment unit that assesses, by a convolutional neural network, the denomination and the position of a lesion which is present in a digestive system endoscopic image of a patient captured by a digestive system endoscopic imaging device and information about accuracies thereof; and a display control unit that performs control for generating an analysis result image in which the denomination and the position of the lesion and the accuracies thereof are displayed and for displaying the image on the digestive system endoscopic image.

Abstract: A robot climbing control method is disclosed. The method obtains an RGB color image and a depth image of stairs, extracts an outline of a target object of a target step on the stairs from the RGB color image, determines relative position information of the robot and the target step according to the depth image and the outline of the target object, and controls the robot to climb the target step according to the relative position information. The embodiment of the present disclosure allows the robot to effectively adjust postures and forward directions on any size of and non-standardized stairs and avoids the deviation of the walking direction, thereby improving the effectiveness and safety of the stair climbing of the robot.

Abstract: An automated documentation intake and processing system involves pre-processing source images of intake documents and applying a YOLO CNN model to identify fields of interest (FOIs) therein which may contain target data. The system classifies FOIs and upsamples the contents in order to digitize and extract target data. The system may be trained to differentiate between target data and non-target data and incorporates an adjustable confidence scores which reflects the system's degree of accuracy at predicting the correct FOI (e.g., name, address, insurance number, vehicle registration number). The system is pre-trained to detect a subset of documentation types, form fields, and form field types. However, the system is configured to adapt to variations of the same types of documentation, such as different insurance cards or driver licenses from different states.

Abstract: A method of feedback controlling a 3D printing process in real time, and a system therefor are disclosed. The method includes collecting big data, generated through 3D printing experiments, related to process variables of 3D printing, measurement signals, and 3D printing quality of the 3D printing object; building an artificial neural network model by performing machine-learning based on the collected big data; evaluating whether or not a 3D printing quality of the 3D printing object is abnormal in real time based on an actual measurement signal of the 3D printing object and the artificial neural network model; and feedback controlling printing quality of the 3D printing object in real time based on the evaluation result of whether or not the 3D printing quality of the 3D printing object is abnormal.

Abstract: A system and a computer-implemented method for validating label data includes receiving the label data and segmenting it into one or more parts using a first machine learning model. Further, from the segmented label data a first plurality of attributes, including text and images, are extracted. The method further includes receiving ground truth data associated with the label data and extracting a second plurality of attributes from the ground truth data. The first and second plurality of attributes are then compared using a second machine learning model and the result of comparison are displayed on a three pane user interface. Further, the label data is validated based on the displayed results.

Abstract: A model is trained to create a probability distribution of counts based on counts of distinct values stored by person profiles in a field. The model is trained to create another probability distribution of counts based on other counts of other distinct values stored by the person profiles in another field. The count of distinct values stored by a person profile in the field is identified. Another count of distinct values stored by the person profile in the other field is identified. A score is determined based on a cumulative distribution function of the count under the probability distribution of counts. Another score is determined based on the cumulative distribution function of the other count under the other probability distribution of counts. If the score and the other score combine in an overall score that satisfies a threshold, a message is output about the person profile being suspected of corruption.

Abstract: Disclosed herein are systems and methods to detect a wide range of eardrum abnormalities by using high-resolution otoscope images and report the condition of the eardrum as “normal” or “abnormal.

Abstract: A non-transitory computer-readable medium includes program code that is stored thereon. The program code is executable by one or more processing devices for performing operations including generating, by a model that includes trainable components, a learned image representation of a target image. The operations further include generating, by a text embedding model, a text embedding of a text query. The text embedding and the learned image representation of the target image are in a same embedding space. Additionally, the operations include generating a class activation map of the target image by, at least, convolving the learned image representation of the target image with the text embedding of the text query. Moreover, the operations include generating an object-segmented image using the class activation map of the target image.

Abstract: A method and a system for diagnosing a plant disease and an insect pest are provided. The method includes: obtaining an plant image; determining a candidate specie and candidate disease and insect pest information corresponding to at least part of the candidate specie according to the plant image when a current diagnosis mode is a passive diagnosis mode; screening out the candidate disease and insect pest information of the candidate specie according to a first preset condition for the candidate specie with the corresponding candidate disease and insect pest information; and outputting at least part of remaining disease and insect pest information after screening out when there is the remaining disease and insect pest information.

Abstract: Systems and methods are described for registering a fingerprint based on optical fingerprint recognition. A fingerprint recognition window is displayed and a fingerprint frequency of a user fingerprint is analyzed based on reflected light received through the fingerprint recognition window using a fingerprint recognition sensor driven by a pixel binning scheme based on a first binning size. The first binning size may be evaluated based on the fingerprint frequency. If the first binning size is not appropriate for registering the user fingerprint, a second binning size is selected. Fingerprint information of the user fingerprint may be obtained again based on using the fingerprint recognition sensor driven by a pixel binning scheme based on the second binning size.

Abstract: The embodiments herein related to a method in a network node, a network node, a method in a user equipment and a user equipment for handling base sequences in a communications network. The network node is configured to communicate with a first user equipment. The network node comprises information about a default base sequence and an alternative base sequence. The network node determines, for the first user equipment, that the alternative base sequence should replace the default base sequence. The network node sends information about the determined alternative base sequence to the first user equipment.

Abstract: Systems and methods for night vision combining sensor image types. Some implementations may include obtaining a long wave infrared image from a long wave infrared sensor; detecting an object in the long wave infrared image; identifying a region of interest associated with the object; adjusting a control parameter of a near infrared sensor based on data associated with the region of interest; obtaining a near infrared image captured using the adjusted control parameter of the near infrared sensor; and determining a classification of the object based on data of the near infrared image associated with the region of interest.

Abstract: A method and an apparatus are provided. The method includes receiving a video with a first plurality of frames having a first resolution; generating a plurality of warped frames from the first plurality of frames based on a first type of motion compensation; generating a second plurality of frames having a second resolution, wherein the second resolution is of higher resolution than the first resolution, wherein each of the second plurality of frames having the second resolution is derived from a subset of the plurality of warped frames using a convolutional network; and generating a third plurality of frames having the second resolution based on a second type of motion compensation, wherein each of the third plurality of frames having the second resolution is derived from a fusing a subset of the second plurality of frames.

Abstract: Methods and systems for determining a ball shot attempt location on a ball court are disclosed. The method includes receiving an input video, where the input video includes the ball shot attempt and one or more players; receiving a declared ball shot attempt; determining a potential shooter by detecting a moving object flow in the input video; determining a potential shooter image region in the image plane by analyzing an overlapping image area of the ball trajectory and the potential shooter; determining a shooter foot-on-court time; determining a most probable pair of shooter legs in the potential shooter image region; determining a shooter foot location based on the most probable pair of shooter legs; and determining a ball shot attempt location on the ball court by applying a camera projection to the shooter foot location.

Abstract: The invention provides for a magnetic resonance imaging system (100) for acquiring magnetic resonance data (142) from a subject (118) within a measurement zone (108). The magnetic resonance imaging system (100) comprises: a processor (130) for controlling the magnetic resonance imaging system (100) and a memory (136) storing machine executable instructions (150, 152, 154), pulse sequence commands (140) and a dictionary (144). The pulse sequence commands (140) are configured for controlling the magnetic resonance imaging system (100) to acquire the magnetic resonance data (142) of multiple steady state free precession (SSFP) states per repetition time. The pulse sequence commands (140) are further configured for controlling the magnetic resonance imaging system (100) to acquire the magnetic resonance data (142) of the multiple steady state free precession (SSFP) states according to a magnetic resonance fingerprinting protocol. The dictionary (144) comprises a plurality of tissue parameter sets.

Abstract: A deep learning-based method for calculating an overhang of a battery includes the following steps: obtaining a training sample image set; training a neural network according to the training sample image set to obtain a segmentation network model; detecting an object detection image of the battery to be detected according to the segmentation network model to obtain a corresponding first binarized image; obtaining top coordinates of each of a positive electrode and a negative electrode of the battery to be detected according to the first binarized image; and calculating the overhang of the battery to be detected according to the top coordinates.

Abstract: The present approach relates to a system capable of life-long learning in a deep learning context. The system includes a deep learning network configured to process an input dataset and perform one or more tasks from among a first set of tasks. As an example, the deep learning network may be part of an imaging system, such as a medical imaging system, or may be used in industrial applications. The system further includes a learning unit communicatively coupled to the deep learning network 102 and configured to modify the deep learning network so as to enable it to perform one or more tasks in a second task list without losing the ability to perform the tasks from the first list.

Abstract: An image processing apparatus, includes a memory; and a processor coupled to the memory and configured to: generate a trained machine learning model by learning a machine learning model using a first set of image data, output an inference result by inputting a second set of image data to the trained machine learning model, and process a region of interest at a time of inference with respect to image data for which an inference result is correct in the second set of image data.

Abstract: A convolutional neural network performs: a first masking process of masking a pixel region not to be inspected, by computing pixel values of corresponding pixels of an inspection image and of a mask image; an intermediate process for extracting a feature image from the inspection image that has been subjected to the first masking process; and a second masking process of masking the pixel region not to be inspected, by computing the pixel values of corresponding pixels of the inspection image that has been subjected to the intermediate process and of the mask image that has been subjected to a process identical to the intermediate process.