Certain aspects of the present disclosure provide techniques for cross-task distillation. A depth map is generated by processing an input image using a first machine learning model, and a segmentation map is generated by processing the depth map using a second machine learning model. A segmentation loss is computed based on the segmentation map and a ground-truth segmentation map, and the first machine learning model is refined based on the segmentation loss.

A method of detecting damages in a freight container, comprising capturing a first image of a part of the freight container at an angle deviating from a perpendicular direction in respect of the part of the container, capturing a second image of the same part of the container at an angle substantially perpendicular in respect of the same part of the container, analysing the first and second images for detecting a damage in the part of the container, and in response to detecting the damage in the part of the container providing a damage information image regarding to the part of the container, the damage information image being an image based on the second image regarding to the respective part of the container and including damages detected in at least one of the respective first image or this second image.

Also, a system for detecting damages in a freight container.

The present disclosure relates to training and utilizing an image exposure transformation network to generate a long-exposure image from a single short-exposure image (e.g., still image). In various embodiments, the image exposure transformation network is trained using adversarial learning, long-exposure ground truth images, and a multi-term loss function. In some embodiments, the image exposure transformation network includes an optical flow prediction network and/or an appearance guided attention network. Trained embodiments of the image exposure transformation network generate realistic long-exposure images from single short-exposure images without additional information.

A grossing station comprising a table and a hood connected to the table. The grossing station also comprises a quick release camera system secured to a surface of the hood, wherein the quick release camera system is detachable from the hood for close up angles and zoomed location still or video images of the body or material being examined on a surface of the grossing station.

A system, apparatus and method are provided for facilitating inspection of a target object. In the context of a method, an image is received that includes the target object. The method applies a first mask to at least a portion of the image in order to mask one or more foreground objects that at least partially block view of the target object. The method further includes applying a second mask to at least a portion of the image in order to mask a background of the target object. The method additionally includes analyzing the image of the target object following application of the first and second masks in order to identify one or more regions of the target object that merit further inspection. A corresponding system and apparatus are also provided.

The present invention relates to a method and system for automatic localisation of static objects in an urban environment. More particularly, the present invention relates to the use of noisy 2-Dimensional (2D) image data to identify and determine 3-Dimensional (3D) positions of objects in large scale urban or city environments. Aspects and/or embodiments seek to provide a method, system, and vehicle for automatically locating static 3D objects in urban environments by using a voting-based triangulation technique. Aspects and/or embodiments also provide a method for updating map data after automatically new 3D static objects in an environment.

Systems and techniques are described for image processing. An imaging system can include an image sensor that captures image data. An image signal processor (ISP) of the imaging system can demosaic the image data. The imaging system can input the image data into one or more trained machine learning models, in some cases along with metadata associated with the image data. The one or more trained machine learning models can output settings for a set of parameters of the ISP based on the image data and/or the metadata. The imaging system can generate an output image by processing the image data using the ISP, with the parameters of the ISP set according to the settings. Each pixel of the pixels of the image data can be processed using a respective setting for adjusting a corresponding parameter. The parameters of the ISP can include gain, offset, gamma, and Gaussian filtering.

An image processing method includes acquiring a captured image obtained by imaging, generating a first image by correcting a blur component of the captured image, and generating a second image based on the captured image, the first image, and weight information. The weight information is generated based on (i) information on brightness of the captured image or information on a scene of the captured image and (ii) information on a saturated area in the captured image.

A neural network system for identifying positions of objects in an input image can include an object detector neural network, a memory interface subsystem, and an external memory. The object detector neural network is configured to, at each time step of multiple successive time steps, (i) receive a first neural network input that represents the input image and a second neural network input that identifies a first set of positions of the input image that have each been classified as showing a respective object of the set of objects, and (ii) process the first and second inputs to generate a set of output scores that each represents a respective likelihood that an object that is not one of the objects shown at any of the positions in the first set of positions is shown at a respective position of the input image that corresponds to the output score.

An image processing device includes a processor; and a memory to store the program which performs processing including: measuring a first texture amount indicating luminance variation in image regions, based on the image regions obtained by dividing the captured image and previously determining an image processing target region based on the first texture amount; judging whether the image processing on a current captured image is necessary based on luminance of the image processing target region in the current captured image; calculating a second texture amount indicating luminance variation in the image processing target region, and judging whether the image processing should be performed on the image processing target region or not based on the second texture amount; and performing the image processing on the image processing target region on which the judging based on the second texture amount is that the image processing should be performed.