A method for optimizing a neural network is provided, including: (1) capturing, via a first sensor group having a first field of view, a first sample set having a first sensor domain corresponding to the first field of view; (2) capturing, via a second sensor group having a second field of view, a second sample set having a second sensor domain corresponding to the second field of view; (3) generating regions of interest of the second sample set; (4) translating the regions of interest to the first sensor domain; (5) identifying nodes of the neural network which correspond to the translated regions; and (6) optimizing the neural network by at least one of (a) increasing the weight value of the nodes corresponding to the one or more translated regions and (b) decreasing the weight value of the nodes not corresponding to the one or more translated regions.

A method for optimizing a neural network is provided, including: (1) capturing, via a first sensor group having a first field of view, a first sample set having a first sensor domain corresponding to the first field of view; (2) capturing, via a second sensor group having a second field of view, a second sample set having a second sensor domain corresponding to the second field of view; (3) generating regions of interest of the second sample set; (4) translating the regions of interest to the first sensor domain; (5) identifying nodes of the neural network which correspond to the translated regions; and (6) optimizing the neural network by at least one of (a) increasing the weight value of the nodes corresponding to the one or more translated regions and (b) decreasing the weight value of the nodes not corresponding to the one or more translated regions.

Techniques related to performing object or person association or correspondence in multi-view video are discussed. Such techniques include determining correspondences at a particular time instance based on separately optimizing correspondence sub-matrices for distance sub-matrices based on two-way minimum distance pairs between frame pairs, generating and fusing tracklets across time instances, and adjusting correspondence, after such tracklet processing, via elimination of outlier object positions and rearrangement of object correspondence.

A method for detecting a field navigation line after ridge sealing of crops includes the following steps. A field crop image is acquired. Image color space transformation, image binaryzation, longitudinal integration, neighborhood setting and region integration calculation are sequentially performed on the field crop image to obtain a crop row image. Detections of an initial middle ridge, a left ridge and a right ridge are performed on the crop row image to obtain center lines of the initial middle ridge, left ridge and right ridge. Center lines of a left (right) crop row are established by using an area 1 between the center lines of the left (right) ridge and the initial middle ridge. A center line model of a middle ridge is established by using an area 0 between the center lines of the left and right crop rows, namely a navigation line of a field operation machine.

A method for image segmentation includes (a) clustering, based upon k-means clustering, pixels of an image into first clusters, (b) outputting a cluster map of the first clusters (c) re-clustering the pixels into a new plurality of non-disjoint pixel-clusters, and (d) classifying the non-disjoint pixel-clusters in categories, according to a user-indicated classification. Another method for image segmentation includes (a) forming a graph with each node of the graph corresponding to a first respective non-disjoint pixel-cluster of the image and connected to each terminal of the graph and to all other nodes corresponding to other respective non-disjoint pixel-clusters that, in the image, are within a neighborhood of the first respective non-disjoint pixel-cluster, (b) setting weights of connections of the graph according to a user-indicated classification in categories respectively associated with the terminals, and (c) segmenting the image into the categories by cutting the graph based upon the weights.

A method for image segmentation includes (a) clustering, based upon k-means clustering, pixels of an image into first clusters, (b) outputting a cluster map of the first clusters (c) re-clustering the pixels into a new plurality of non-disjoint pixel-clusters, and (d) classifying the non-disjoint pixel-clusters in categories, according to a user-indicated classification. Another method for image segmentation includes (a) forming a graph with each node of the graph corresponding to a first respective non-disjoint pixel-cluster of the image and connected to each terminal of the graph and to all other nodes corresponding to other respective non-disjoint pixel-clusters that, in the image, are within a neighborhood of the first respective non-disjoint pixel-cluster, (b) setting weights of connections of the graph according to a user-indicated classification in categories respectively associated with the terminals, and (c) segmenting the image into the categories by cutting the graph based upon the weights.

A method for automatically identifying global solar photovoltaic (PV) panels based on a cloud platform by using remote sensing. Optical images in a study area for a whole specific year are collected based on the cloud platform, and preprocessing is performed to obtain a surface reflectance image. Seven time-series images are derived and constructed based on spectral features of a solar PV panel: a solar PV panel index image, a water index image, a vegetation index image, a difference image between a first shortwave infrared band and a second shortwave infrared band, a difference image between the first shortwave infrared band and a near-infrared band, a blue band image, and a first shortwave infrared band image. Data in the seven time-series images are synthesized and reconstructed to obtain input data required by a model. A remote sensing theoretical model for automatically identifying a solar PV panel is constructed.

The present disclosure provides an object tracking method and an object tracking device. The method includes: acquiring a human-face region of an image frame so as to determine a human-body region; extracting a human-body feature from the human-body region, and determining whether a plurality of historical object trajectories match the human-body feature; in response to that one of the plurality of historical object trajectories matches the human-body feature, updating an age of the human-body feature to a preset value; and in response to that none of the plurality of historical object trajectories matches the human-body feature, adding an object trajectory corresponding to the human-body feature to the plurality of historical object trajectories. Thus, a better tracking effect may be achieved.

The present disclosure provides an object tracking method and an object tracking device. The method includes: acquiring a human-face region of an image frame so as to determine a human-body region; extracting a human-body feature from the human-body region, and determining whether a plurality of historical object trajectories match the human-body feature; in response to that one of the plurality of historical object trajectories matches the human-body feature, updating an age of the human-body feature to a preset value; and in response to that none of the plurality of historical object trajectories matches the human-body feature, adding an object trajectory corresponding to the human-body feature to the plurality of historical object trajectories. Thus, a better tracking effect may be achieved.

There is provided an image processing device that processes a projection image presented to a plurality of persons at the same time. The image processing device specifies an overlapping area in which fields of view of two or more users overlap based on information on each user, classifies objects included in the overlapping area into a first object group and a second object group, generates a common image common to all users, made up of the first object group, generates individual images different for each user, made up of the second object group, and determines an output protocol for displaying the individual images.