An electronic device that specifies or determines information associated with an application is described. During operation, the electronic device may identify one or more objects of interest in an acquired image. Then, the electronic device may determine or specify classifications (such as names) for the one or more objects of interest. Moreover, the electronic device may analyze a context of the one or more objects of interest in order to determine one or more inspection criteria. Once the one or more inspection criteria are determined, the electronic device may receive publishing information (such as designated recipients) and privacy settings (such as is the application private or public). Next, the electronic device may generate the application using the specified or determined information, and may publish or provide the application to one or more other electronic devices associated with the designated recipients using the publishing information and the privacy settings.

According to one implementation, a quality control (QC) system for annotated content includes a computing platform having a hardware processor and a system memory storing an annotation culling software code. The hardware processor executes the annotation culling software code to receive multiple content sets annotated by an automated content classification engine, and obtain evaluations of the annotations applied by the automated content classification engine to the content sets. The hardware processor further executes the annotation culling software code to identify a sample size of the content sets for automated QC analysis of the annotations applied by the automated content classification engine, and cull the annotations applied by the automated content classification engine based on the evaluations when the number of annotated content sets equals the identified sample size.

An event classification is trained by machine learning. An anomaly detection for detecting events in an image data set is thereby performed. Based on the performance of the anomaly detection, a model assumption of the event classification is determined. An image data set may include a plurality of images, and each image may include an array of pixels. Further, an image data set may include volume data and/or a time sequence of images and in this way represent a video sequence.

Aspects of the disclosure relate to training a labeling model to automatically generate labels for objects detected in a vehicle's environment. In this regard, one or more computing devices may receive sensor data corresponding to a series of frames perceived by the vehicle, each frame being captured at a different time point during a trip of the vehicle. The computing devices may also receive bounding boxes generated by a first labeling model for objects detected in the series of frames. The computing devices may receive user inputs including an adjustment to at least one of the bounding boxes, the adjustment corrects a displacement of the at least one of the bounding boxes caused by a sensing inaccuracy. The computing devices may train a second labeling model using the sensor data, the bounding boxes, and the adjustment to increase accuracy of the second labeling model when automatically generating bounding boxes.

An automated airfield ground lighting inspection system and method is disclosed. An image acquisition means captures image streams of the airfield ground lighting system lights when moved across an airfield. A location sensor detects positional information for the image acquisition means when capturing the plurality of images comprising the image streams. An image processor coupled to the image acquisition means and the location sensor processes the image stream of a light of the airfield ground lighting system by: (a) associating characteristics of a plurality of points in an image with an item in the light to be checked, and using this association for extraction of the points; (b) verifying each extracted point; and (c) determining the state of the light of the image stream by processing the verified extracted points comprising an item to be checked.

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.

An inspection apparatus includes: an image capturing apparatus configured to capture an image of an object to be inspected: a determination unit configured to determine, based on the image, whether or not the object to be inspected includes a defect, using an identification device that has been trained using learning data: an input unit configured to accept an input indicating whether or not a determination result by the determination unit is correct; an extraction unit configured to extract a partial image of the image based on which the determination has been made; and a generation unit configured to generate new learning data based on the partial image, if a fact that the determination result by the determination unit is not correct has been input.

A method for generating training data for a deep learning network is provided. The method includes steps of: an auto labeling device (a) (i) allowing a labeling network to label acquired test images and generate primary bounding boxes, primary class information, and primary confidence scores, (ii) allowing a labeler to verify labeled primary objects and generate correction-related class information, and (iii) setting first and second threshold confidence scores; (b) (i) allowing the labeling network to label acquired unlabeled images and generate secondary bounding boxes, secondary class information, and secondary confidence scores, (ii) allowing an object difficulty estimation module to generate object difficulty scores and determine object difficulty classes, and (iii) allowing an image difficulty estimation module to determine image difficulty classes; and (c) allowing the labeler to verify the first labeled images and generating the training data comprised of second labeled images and the verified first labeled images.

A first and second scoring endpoint with payload logging are deployed. At the second scoring endpoint, native data and a user-generated score for the native data are received, the native data is pre-processed into readable data for the deep-learning model, and the user-generated score and the readable data are output to the first scoring endpoint, which is associated directly with the deep-learning model. A raw payload that includes the native data is output to a payload store. At the first scoring endpoint, the readable data and the user-generated score are processed by the deep-learning model, which outputs a transformed payload and a prediction, respectively, to the payload store. The raw payload is matched with the transformed payload and the prediction to produce a comprehensive data set, which is evaluated to describe a set of transformation parameters. The deep-learning model is retrained to account for the set of transformation parameters.

An image capturing system includes a camera unit capable of performing first image capturing to capture an image of a cell before transfer work is performed and second image capturing to capture the image of the cell after the transfer work; a determination unit configured to make a first determination to determine whether to select the cell based on selection criterion data from the image captured by the first image capturing and a second determination to determine whether to select the cell from the image captured by the second image capturing; a storage unit configured to store the selection criterion data; and a correction unit configured to update the selection criterion data stored in the storage unit to make the first determination and the second determination about the cell identical to each other in a subsequent determination when the first determination and the second determination have different determination results.