Systems and methods are provided for classifying microbial cells according to morphological features of microcolonies. A dark-field objective is employed to acquire a dark-field image of a microcolony during a microcolony growth phase that is characterized by phenotypic expression of microcolony morphological features which evolve with time and are differentiated among classes of microbial cell types. The dark-field image is processed to classify the microcolony according to two or more microbial cell types, such as Gram status and/or speciation. The dark-field objective may have a numerical aperture selected to facilitate the imaging of microcolony morphological features, residing, for example, between 0.15 and 0.35. A set of dark-field images of a microcolony may be collected during the microcolony growth phase and processed to classify the microcolony. Classification may be performed according to a temporal ordering of the dark-field images, for example, using a recurrent neural network.

The present embodiment provides an image restoration method and apparatus which generate independent different restoration models by performing learning for each of different resolutions, receive a distorted image, and apply a restoration model corresponding to the resolution of the distorted image among the independent different restoration models to restore the distorted image into an improved upscaled image centering on a restoration target object within the distorted image.

A method of producing information from at least one camera image of an object, including: A) recording raw image data of the at least one camera image, B) evaluating the raw image data by a mathematical linkage to produce combination image data, C) deriving the information from the combination image data, D) outputting the information, E) determining an actual measure for a data quality of the raw image data prior to or after evaluation steps in step B), F) determining a deviation between the actual measure for the data quality and a target measure for the data quality of the raw image data of at least one camera image, and G) again recording all raw image data of those camera images, for which the deviation determined in step F) is greater than a predetermined threshold value and repeating at least one evaluation step from step B) and steps C) to F) either until the deviation determined in step F) for the raw image data of all camera images from the plurality of camera images is less than the threshold value or until a predetermined termination condition is fulfilled.

A binary image is generated from a multivalued image based on a threshold, an edge is extracted from the multivalued image, an edge image is generated by correcting a position of the extracted edge, and a synthetic binary image is generated by synthesizing the edge image and the binary image.

Super-resolution label-free microscopy is provided using multiplexed, temporally modulated acquisition patterns of emission point spread functions (“PSFs”). Supercontinuum ultrafast pulses can be used to enhance nonlinear processes, such as autofluorescence and harmonic generation, in order to provide super-resolution imaging of nonlinear label-free signals. Images can be reconstructed using various reconstruction techniques, including pixel reassignment, wavelet reconstruction, and deep learning model-based reconstructions.

The present disclosure relates to systems, methods, and non-transitory computer readable media that utilize a continuous kernel neural network that learns continuous reconstruction kernels to merge digital image samples in local neighborhoods and generate enhanced digital images from a plurality of burst digital images. For example, the disclosed systems can utilize an alignment model to align image samples from burst digital images to a common coordinate system (e.g., without resampling). In some embodiments, the disclosed systems generate localized latent vector representations of kernel neighborhoods and determines continuous displacement vectors between the image samples and output pixels of the enhanced digital image. The disclosed systems can utilize the continuous kernel network together with the latent vector representations and continuous displacement vectors to generated learned kernel weights for combining the image samples and generating an enhanced digital image.

A computer-implemented method of generating training data to be used to train a machine learning model for generating a segmentation mask of an image containing overlapping particles. Training data is generated from sparse particle images which contain no overlaps. Generating masks for non-overlapping particles is generally not a problem if the particles can be identified clearly; in many cases simple methods such as thresholding already yield usable masks. The sparse images can then be combined to images which contain artificial overlaps. The same can be done for the masks as well which yields a large amount of training data, because of the many combinations which can be created from just a small set of images. The method is simple yet effective and can be adapted to many domains for example by adding style-transfer to the generated images or by including additional augmentation steps.

Systems, methods, storage media, and computing platforms for scanning items at the point of manufacturing are disclosed. Exemplary implementations may: receive a first set of images of an item from a first set of camera sources; detect a code in the first set of images; combine, responsive to detecting the code, along a second axis perpendicular to the first axis, the first set of images into a first set of combined images; rotate parallel to the first axis; and combine along the first axis.

Systems, devices, methods, and computer processing products for automatically checking for errors in segmentation (contouring) using heuristic and/or statistical evaluation methods.

A method for generating a 2D orthomosaic map including obtaining a series of images of a field from a camera located on a ground based vehicle, processing the series of images to mark pixels of the ground based vehicle and optionally an implement, identifying, marking, and removing pixels containing plants, stitching together the series of images into a single map, and reintroducing pixels containing plants into the single map.