Systems and methods for image tagging are described. In some embodiments, images with problematic tags are identified after applying an auto-tagger. The images with problematic tags are then sent to an object detection network. In some cases, the object detection network is trained using a training set selected to improve detection of objects associated with the problematic tags. The output of the object detection network can be merged with the output of the auto-tagger to provide a combined image tagging output. In some cases, the output of the object detection network also includes a bounding box, which can be used to crop the image around a relevant object so that the auto-tagger can be reapplied to a portion of the image.

Systems and methods for tracking inventory items in an area of real space are disclosed. The method includes receiving a signal generated in dependence on sensors. The signal indicates a change to a portion of an image of an area of real space. The method includes, in response to receiving the signal, implementing a trained location detection model to determine, based on inputs, whether an inventory item identified in the portion of the image has changed a position in the area of real space. The method includes implementing a trained item classification model to determine a classification of the inventory item. The method includes updating an inventory database with inventory item data determined in dependence on the classification of the inventory item to provide an updated map of the area of real space as a result of the received signal indicating the change to the portion of the image.

An intelligent zooming method and an electronic device using the same are provided. The intelligent zooming method includes the following steps. A text paragraph corresponding to the text is merged. The text paragraph is automatically arranged according to the text paragraph and a text magnification box. The text paragraph in the text magnification box is enlarged according to a text magnification ratio. A block group containing the block and other blocks connected thereto is merged. A block magnification ratio is adjusted according to the block group and a block magnification box. The block group in the block magnification box is enlarged according to the block magnification ratio. The picture is cropped to obtain an object. A picture magnification ratio is adjusted according to the object and a picture magnification box. The object in the picture magnification box is enlarged according to the picture magnification ratio.

Described herein are systems and methods of converting media dimensions. A device may identify a set of frames from a video in a first orientation as belonging to a scene. The device may receive a selected coordinate on a frame of the set of frames for the scene. The device may identify a first region within the frame including a first feature corresponding to the selected coordinate and a second region within the frame including a second feature. The device may generate a first score for the first feature and a second score for the second feature. The first score may be greater than the second score based on the first feature corresponding to the selected coordinate. The device may crop the frame to include the first region and the second region within a predetermined display area comprising a subset of regions of the frame in a second orientation.

An automatic abnormal cell recognition method, the method including: 1) scanning a slide using a digital pathological scanner and obtaining a cytological slide image; 2) obtaining a set of centroid coordinates of all nuclei that is denoted as CentroidOfNucleus by automatically localizing nuclei of all cells in the cytological slide image using a feature fusion based localizing method; 3) obtaining a set of cell square region of interest (ROI) images that are denoted as ROI_images; 4) grouping all cell images in the ROI_images into different groups based on sampling without replacement, where each group contains ROW×COLUMN cell images with preset ROW and COLUMN parameters; obtaining a set of splice images; and 5) classifying all cell images in the splice image simultaneously by using the splice image as an input of a trained deep neural network; and recognizing cells classified as abnormal categories.

A workpiece surface defect detection device or the like that is capable of stably detecting a small surface defect with high accuracy is provided. A plurality of images indicating a portion to be measured of a workpiece serving as a target of detection of a surface defect is obtained in a state where a bright-and-dark pattern of an illumination device is moved relative to the workpiece, and a tentative defect candidate is extracted. When among a plurality of images from which the tentative defect candidate has been extracted, the number of images including the tentative defect candidate is greater than or equal to a threshold that has been set in advance, the tentative defect candidate is determined as a defect candidate. A plurality of images including the determined defect candidate is combined to generate a composite image, and a defect is detected on the basis of the generated composite image.

Embodiments of the present disclosure generally relate to livestreaming methods and systems, and more particularly to whiteboard presentation systems that can be used in a livestreaming or video conferencing environment. In some embodiments, the whiteboard presentation system is configured to perform one or more processing operations, such as capture images on a whiteboard, perform image processing routines on the captured images, and transmit processed images as a video feed to one or more remote users, such as people or locations attending a video conference. The image processing routines can include one or more operations such as image denoising, contrast enhancement, color reconstruction, segmentation of a presenter, and image reconstruction.

The present application relates to a code acquisition system and method. The code acquisition system comprises a transfer track; an image capture device and an image processing device. The image capture device is configured to capture memory stick images. The code area definition module in the image processing device is configured to divide into code areas according to the types, sizes and positions of pre-loaded memory stick codes. The code identification module identifies the memory stick codes according to the memory stick images and the code areas.

A system and method for calibrating a machine vision system on the undercarriage of a rail vehicle while the rail vehicle is in the field is presented. The system enables operators to calibrate the machine vision system without having to remove the machine vision system from the undercarriage of the rail vehicle. The system can capture, by a camera of an image capturing module, a first image of a target. The image capturing module and a drum can be attached to a fixture and the target can be attached to the drum. The system can also determine a number of lateral pixels in a lateral pitch distance of the image of the target, determining a lateral object pixel size based on the number of lateral pixels, and determining a drum encoder rate based on the lateral object pixel size. The drum encoder rate can be programmed into a drum encoder.

Disclosed herein is an object detection system, including apparatuses and methods for object detection. An implementation may include receiving a first image frame from an ROI detection model that generated a first ROI boundary around a first object detected in the first image frame and subsequently receiving a second image frame. The implementation further includes predicting, using an ROI tracking model, that the first ROI boundary will be present in the second image frame and then detecting whether the first ROI boundary is in fact present in the second image frame. The implementation includes determining that the second image frame should be added to a training dataset for the ROI detection model when detecting that the ROI detection model did not generate the first ROI boundary in the second image frame as predicted and re-training the ROI detection model using the training dataset.