Abstract: Disclosed herein are apparatuses and methods for auto-configuring a region of interest (ROI) associated with a camera. In one implementation, A method includes tracking, over a plurality of image frames spanning a first period of time, a plurality of objects moving in an environment, and determining a plurality of respective trajectories of the plurality of objects. The method includes identifying a common trajectory shared by more than one object of the plurality of objects based on comparing each of the plurality of respective trajectories to one another and identifying an ROI that encompasses the common trajectory. The method further includes tracking, over a plurality of new image frames spanning a second period of time, another plurality of objects moving in the environment, and generating an alert that the common trajectory has changed based on new trajectories of the another plurality of objects.

Abstract: A method for detecting information about at least one object and/or at least a part of the free space in a representation (4) of the environment of a system. The method comprises: a) performing ground truth generation; b) performing an obstacle and clearance detection; c) performing a generalization across different cameras and/or different digital image representations.

Abstract: The partial image generation means generates partial images from the input image, and the partial image selection means selects a plurality of training partial images to be used as training data from all the generated partial images. The training means trains a prediction model for predicting a probability that a predetermined feature is included in the selected training partial images. The prediction means performs prediction for all the partial images using the trained prediction model. Then, the partial image selection means selects the plurality of training partial images to be used in a next training based on predicted values for all the partial images. The training of the prediction model by the training means is repeated while the training partial images are updated.

Abstract: Systems and methods for enhanced end-to-end three-dimensional (3-D) object detection are disclosed that improve detecting objects in the 3D space from images, such as monocular camera image that may be captured during the operation the autonomous vehicles. For example, a vehicle can include a processor device detecting one or more objects in a 3D space by predicting 3D bounding boxes and predicting dense depth associated with target assignments. The target assignments correspond to the location of objects within an image of the 3D space of a surrounding environment for the vehicle. The vehicle can also include a controller device that receives the detection of the objects in the 3D space from the processor device and performs autonomous operations. The end-to-end 3D object detection techniques achieve a high level of object detection accuracy, with significant improvements compared to previous methods, due to the simplicity and optimization of its end-to-end functionality.

Abstract: A locating method and apparatus for a robot, and a computer-readable storage medium. The locating method includes: determining current possible pose information of the robot according to current ranging data collected by a ranging unit; determining, according to first current image data collected by an image collection unit, first historical image data matching with the first current image data, the first historical image data being collected by the image collection unit at a historical moment; obtaining first historical pose information of the robot at a moment when the first historical image data is collected; and in response to quantity of the current possible pose information being at least two pieces, matching the first historical pose information with each piece of the current possible pose information, and using matched current possible pose information as current target pose information.

Abstract: A depth estimation system to perform operations that include: receiving image data generated by a client device, the image data comprising a depiction of an environment; identifying a set of image features based on the image data; determining a pose of the client device based on the set of features; generating a depth estimation based on the image data and the pose of the client device; and generating a mesh model of the environment based on the depth estimation.

Abstract: Provided in the embodiments of the present application are a target object detection method and apparatus, and an electronic device and a computer storage medium. The method comprises: when it is determined that a robot arm picks up a target object from any turnover box and the height of the target object is greater than the height of the turnover box, performing target tracking on the target object to obtain real-time position information of the target object; and when it is determined, according to the real-time position information of the target object, that the target object falls from the robot arm, determining an area in which the target object is located according to position information of the target object at the current moment and area position information, wherein the area position information comprises areas in which various types of turnover boxes are located.

Abstract: Embodiments of the inventive concept provide a substrate treating apparatus and a substrate treating method for allowing a position of a substrate to be known with only an analysis of an image with respect to a light source irradiated to a surrounding of the substrate. The inventive concept provides a substrate treating apparatus. The substrate treating apparatus includes a support unit configured to seat a substrate; a first light irradiation unit configured to irradiate a first light source to the substrate; an image acquisition unit configured to acquire an image information with respect to the first light source irradiated to the substrate; and an image analysis unit configured to detect a position of the substrate by inputting the image information and analyzing the first light source irradiated to the substrate within the image information which is input.

Abstract: An information processing apparatus includes at least one memory storing a program, and at least one processor which, by executing the program, causes the information processing apparatus to acquire learning image data, ground truth region image data which indicates a ground truth region of the first region included in the learning image data, and ground truth data on the recognition, acquire estimated region image data which indicates an estimated region of the first region in the learning image data, and train the learning model using a first training dataset constituted of the learning image data, the ground truth region image data and the ground truth data, and a second training dataset constituted of the learning image data, the estimated region image data and the ground truth data.

Abstract: A method of and system are disclosed for detecting various gestures by an imaging device and executing a task assigned to these gestures. For instance, the task could include a number of smart home functions, as well as various text and other communication transmissions. The imaging device, which can be a digital camera, is wirelessly communicable with one or more base stations and/or access points over a wireless local area network. The imaging device transmits images and/or videos, which can be reviewed by a computer vision program (“CV”). CV can identify potential gestures in the images and/or videos and compare any such images and/or videos to a database of known gestures. When a gesture is detected, a corresponding task, such as a smart home action, may be executed. Additionally, CV may use facial recognition to first identify an individual, and then execute a task specific to the identified individual based on a detected gesture.

Abstract: A computer includes a processor and a memory, and the memory stores instructions executable by the processor to determine a quantity of relative motion between a host vehicle and a target vehicle; in response to the quantity of relative motion being below a threshold, determine a pose of the host vehicle subject to a constraint on motion of the host vehicle; and in response to the quantity of relative motion being above the threshold, determine the pose of the host vehicle without the constraint.

Abstract: According to the vision sensor, the operating method of the vision sensor, and the image processing device according to the inventive concepts, event data may be generated in response to the movement of an object, a plurality of pieces of comparison data may be generated by comparing each of a plurality of pieces of event data respectively corresponding to a plurality of sub regions with reference data, a motion count value may be updated by comparing each of the plurality of pieces of event data respectively corresponding to the plurality of sub regions with a first threshold value, and whether an object moves may be determined by comparing an updated final motion count value with a second threshold value.

Abstract: An autonomous vehicle can include a horn; a sensor(s) configured to capture images; and one or more processors. The one or more processors can be configured to receive a sequence of images from the sensor(s), the sequence of images captured by the sensor(s) as the autonomous vehicle was moving; execute a machine learning model using the sequence of images as input to detect a human inside a second vehicle or in the surrounding environment of the autonomous vehicle depicted within the sequence of images; determine, based on the detection of the human inside the second vehicle or in the surrounding environment of the autonomous vehicle within the sequence of images, the human is depicted performing a defined arm gesture within the sequence of images; and activate the horn responsive to the determination that the human is depicted performing the defined arm gesture within the sequence of images.

Abstract: A cell alignment metric detection method, a controller, a detection system and a storage medium are disclosed. The cell alignment metric detection method includes: acquiring a first cell image; inputting the first cell image into a first training model for detection processing to obtain a region of interest; obtaining a second cell image and terminal position information of a second electrode plate according to the region of interest and the first cell image; inputting the second cell image into a second training model for separation processing to obtain a third cell image and terminal position information of a first electrode plate; performing image extraction and analysis on the third cell image to obtain a misalignment quantity; and obtaining the cell alignment metric according to the terminal position information of the second electrode plate and the first electrode plate and the misalignment quantity.

Abstract: There is provided a method and system for 2D motion tracking of a subject engaging in physical exercise using a 2D camera. An image including a portion of a body of the subject is received and a first and a second point on the body are detected. A reference distance between the first and second point is determined, and a reference ratio is calculated based on the reference distance and at least one of the height and the width of the image. The first and second point are mapped into virtual space based on the reference ratio to obtain a first and second virtual point. A movement of at least one of the first and second point is detected in another image and mapped to a movement of a corresponding one of the first and second virtual point in virtual space which is output on a display interface.

Abstract: An apparatus and method with tracking a target are provided. A method includes determining whether data augmentation is to be used to augment a target tracking process, based on determining that data augmentation is to be used, performing the target tracking process based on an augmented image area obtained by the data augmentation on an image area, and outputting a tracking result generated by the target tracking process.

Abstract: A velocity measurement method, system, device and apparatus and storage medium, as well as a velocity field measurement method and system are provided. The velocity measurement method includes: inputting a single-particle image to a trained neural network model to obtain a predicted result image; a pixel position of a particle in the predicted result image is a predicted position of a particle in the single-particle image; the particle in the single-particle image is configured to characterize one particle in a two-dimensional (2D) flow field measurement region; the trained neural network model is trained by taking a previous frame of particle image in a particle image pair as an input and a subsequent frame of particle image as an output true value; and determining a predicted velocity at the pixel position of the particle in the single-particle image.

Abstract: An intelligent method for efficiently classifying concrete cracks from large amounts of image data is proposed, named inverted residual (IR) 7-Efficient Channel Attention and Convolutional Block Attention Module (EC) network. The IR7-EC network consists of a convolutional layer, seven inverted residual-ECA structures, a CBAM attention mechanism, a pooling layer, and multiple fully connected layers that are sequentially connected. The inverted residual-ECA structure consists of two components: a depthwise separable convolution-based inverted residual structure and an ECA attention mechanism. The new inverted residual structure facilitates the feature extraction of concrete cracks. Compared to conventional network structures like VGG and Resnet, the proposed IR7-EC network excels in both accuracy and efficiency. Once the IR7-EC network is fully trained, it can accurately classify various types of concrete cracks in captured images.

Abstract: What is disclosed is methods and systems for monitoring micro-object movements. Images depicting micro-objects moving across a backplane is received. Based on processing by a machine learning based image processing model, a first location and orientation of each micro-object in a first image and a second location and orientation of each micro-object in a second image is determined. Based on estimates of forces applied during a time between capture of the first and second images, movement during the time of a micro-object in the first image is estimated. A likely image of the micro-object in the second image is inferred based on the likely image having a location and orientation in the second image that corresponds to a combination of the movement and the location and orientation of the micro-object in the first image. An indication that the micro-object in the first image corresponds to the likely image is output.

Abstract: A method of product detection includes receiving, at a product detector from a product source, a first annotation package for a first product and a second annotation package for a second product. An artificial intelligence model is trained to detect the first product based on the first annotation package and the second product based on the second annotation package. The artificial intelligence model is implemented on the product detector. At the product detector, the first product is categorized into a first category of products and the second product is categorized into a second category of products. A subscription is received from a retailer to one of: the first category of products; the second category of products; and the first category and the second category of products. An image is received at the product detector from the retailer. The first product is detected in the image by the artificial intelligence model.