Embodiments of the present invention provide a system that can be used to classify a feedback image in a user review into a semantically meaningful class. During operation, the system analyzes the captions of feedback images in a set of user reviews and determines a set of training labels from the captions. The system then trains an image classifier with the set of training labels and the feedback images. Subsequently, the system generates a signature for a respective feedback image in a new set of user reviews using the image classifier. The signature indicates a likelihood of the image matching a respective label in the set of training labels. Based on the signature, the system can allocate the image to an image cluster.

A method for training and operating an autonomous vehicle. The method includes operating the autonomous vehicle with a control module. The control module includes a series of sensors configured to detect objects or situations in a path of the autonomous vehicle, and a machine learning algorithm trained to classify the objects or interpret the situations detected by the sensors. The method also includes prompting a safety driver of the autonomous vehicle to provide a response when the machine learning algorithm is unable to classify one of the objects or is unable to interpret one of the situations. The method further includes receiving, at the control module, the response from the safety driver, and providing the response from the safety driver as additional training data to the machine learning algorithm.

A human expert may initially label a white light image of teeth, and computer vision may initially label a filtered fluorescent image of the same teeth. Each label may indicate presence or absence of dental plaque at a pixel. The images may be registered. For each pixel of the registered images, a union label may be calculated, which is the union of the expert label and computer vision label. The union labels may be applied to the white light image. This process may be repeated to create a training set of union-labeled white light images. A classifier may be trained on this training set. Once trained, the classifier may classify a previously unseen white light image, by predicting union labels for that image. Alternatively, the items that are initially labeled may comprise images captured by two different imaging modalities, or may comprise different types of sensor measurements.

A system includes a memory; and a processor configured to train a first machine learning model based on the first dataset labeling; provide the second dataset to the trained first machine learning model to generate an updated second dataset including an updated second dataset labeling, determine a first difference between the updated second dataset labeling and the second dataset labeling; train a second machine learning model based on the updated second dataset labeling if the first difference is greater than a first threshold value; provide the first dataset to the trained second machine learning model to generate an updated first dataset including an updated first dataset labeling, determine a second difference between the updated first dataset labeling and the first dataset labeling; and train the first machine learning model based on the updated first dataset labeling if the second difference is greater than a second threshold value.

Aspects of the disclosure relate to determining a sign type of an unfamiliar sign. The system may include one or more processors. The one or more processors may be configured to receive an image and identify image data corresponding to a traffic sign in the image. The image data corresponding to the traffic sign may be input in a sign type model. The processors may determine that the sign type model was unable to identify a type of the traffic sign and determine one or more attributes of the traffic sign. The one or more attributes of the traffic sign may be compared to known attributes of other traffic signs and based on this comparison, a sign type of the traffic sign may be determined. The vehicle may be controlled in an autonomous driving mode based on the sign type of the traffic sign.

The invention relates to a device that receives images from one or more cameras, process images and automatically detects unknown humans in the field of view of the camera, for example to prevent burglary. In order to do so, the device comprises a processing logic configured to detect faces, recognize faces and verify if a face corresponds to a face in a collection of faces of known humans. If a face is detected, but does not correspond to a known face, an alarm event is triggered. The processing logic is further configured to classify objects in the image in classes of object comprising at least a human class. If a human is recognized, but no face has been detected for this human, an alarm event is also triggered. Thus, an alarm can be triggered in any case wherein a human is detected, which is not a known, trusty human.

A machine continuous learning method with a neural network object classifying function is applied to a monitoring camera apparatus having a processor with an object classifier. The machine continuous learning method includes utilizing the processor to receive an image, utilizing the object classifier to analyze the image for generating a first parameter and a second parameter, utilizing the processor to determine whether the first parameter belongs to at least one cluster established by human feedback, and utilizing the processor to output a label of the at least one cluster or the second parameter generated by the object classifier according to a determination result.

A medical registration training component executing within a medical registration system performs a training medical registration operation on a pair of medical studies. Responsive to the medical registration training system determining that the training medical registration operation succeeds, the medical registration training system records a medical registration instance for the pair of medical studies in a medical registration history and marks the medical registration instance as a positive instance in the medical registration history. Responsive to the medical registration training system determining that the training medical registration operation requires correction, the medical registration training system records a medical registration instance for the pair of medical studies in the medical registration history and marks the medical registration instance as a negative instance in the medical registration history. The medical registration training system trains a failure prediction machine learning model based on the medical registration history using machine learning such that the failure prediction machine learning model predicts whether a new medical registration operation will require correction. Responsive to the failure prediction machine learning model predicting that the new medical registration operation will require correction, the mechanism takes steps to automatically correct the new medical registration operation.

A Best of Burst Selector, or BoB Selector, automatically selects a subjectively best image from a single set of images of a scene captured in a burst or continuous capture mode, captured as a video sequence, or captured as multiple images of the scene over any arbitrary period of time and any arbitrary timing between images. This set of images is referred to as a burst set. Selection of the subjectively best image is achieved in real-time by applying a machine-learned model to the burst set. The machine-learned model of the BoB Selector is trained to select one or more subjectively best images from the burst set in a way that closely emulates human selection based on subjective subtleties of human preferences. Images automatically selected by the BoB Selector are presented to a user or saved for further processing.

A method is provided for reducing a computational cost of deep reinforcement learning using an input image to provide a filtered output image composed of pixels. The method includes generating a moving gate in which the pixels of the filtered output image to be masked are assigned a first gate value and the pixels of the filtered output image to be passed through are assigned a second gate value. The method further includes applying the input image and the moving gate to a GCNN to provide the filtered output image such that only the pixels of the input image used to compute the pixels assigned the second gate value are processed by the GCNN while bypassing the pixels of the input image useable to compute the pixels assigned the first gate to reduce an overall processing time of the input image in order to provide the filtered output image.