Devices and methods for identifying an object in an image
Inventors
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-10546389-B2
Publication Date
2020-01-28
Expiration Date
Abstract
Methods, devices, and computer-readable storage media for identifying an object in an image, the method including using a first neural network to identify an approximate position of an object of interest in an image and identifying based on the approximate position, a section of the image that includes the object of interest. A plurality of sub-images corresponding to the identified section of the image are applied to a plurality of second neural networks to determine a plurality of second positions of the object of interest. The plurality of second positions are statistically analyzed to determine an output position of the object of interest.
Core Innovation
The invention relates to an image object detection system that identifies an object of interest in an image by estimating an approximate position with a first neural network. Based on the identified approximate position, a section of the received image that includes the object of interest is identified, and a plurality of sub-images corresponding to the identified section of the received image is obtained and generated.
Each sub-image is applied to one or more second neural networks, where each second neural network determines a respective position of the object of interest. The second neural networks determine a respective plurality of second positions of the object of interest, and the plurality of second positions are statistically analyzed to determine an output position of the object of interest.
The statistical analysis includes post-processing of second neural network results using bounding rectangles and statistical filtering. Average top, left, right, and bottom sides are determined from bounding rectangles to construct an average rectangle, a center is calculated and translated to a position in the received image, and bounding-rectangle sides that exceed a standard deviation threshold are removed before averaging.
Claims Coverage
The document includes three independent claims that cover a computer-implemented method, a device, and a non-transitory computer-readable medium. Across these claims, the core coverage consists of a first neural network for approximate position, extraction of an image section, generation of multiple sub-images, application of multiple second neural networks to obtain multiple refined positions, and statistical analysis to output a final object position.
Approximate position to image section extraction and multiple sub-image inference
using a first neural network to identify an approximate position of an object of interest in the received image; identifying, based on the identified approximate position, a section of the received image that includes the object of interest; obtaining, by the one or more processors, a plurality of sub-images corresponding to the identified section of the received image
Multiple second neural networks for respective object positions
applying the plurality of sub-images to a plurality of second neural networks, each of the second neural networks determining a respective position of the object of interest, such that the plurality of second neural networks determine a respective plurality of second positions of the object of interest
Statistical analysis to determine an output position
statistically analyzing the plurality of second positions to determine an output position of the object of interest
Together, the independent claims consistently require the same pipeline: approximate position from a first neural network, image-section determination and generation of a plurality of sub-images, multiple second neural networks producing multiple second positions, and statistical analysis producing the output position.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Medical image modalities are explicitly listed as angiography, MRI, CT, ultrasound, PET, SPECT, X-ray, and fluoroscopy, as used within an image object detection system for medical images.
Interested in licensing this patent?