Recognizing fine-grained objects in surveillance camera images
AssigneeNEC Laboratories America, Inc.
InventorsYi-Hsuan Tsai, Manmohan Chandraker, Shuyang Dai, Kihyuk Sohn
Life of this patent
Estimating the term…
Abstract
Systems and methods for recognizing fine-grained objects are provided. The system divides unlabeled training data from a target domain into two or more target subdomains using an attribute annotation. The system ranks the target subdomains based on a similarity to the source domain. The system applies multiple domain discriminators between each of the target subdomains and a mixture of the source domain and preceding target domains. The system recognizes, using the multiple domain discriminators for the target domain, fine-grained objects.
Claims20
A method for recognizing fine-grained objects, comprising:
- dividing, by a processor device, unlabeled training data from a target domain into two or more target subdomains using at least one attribute annotation;
- ranking the two or more target subdomains based on a similarity to the source domain;
- applying multiple domain discriminators between each of the two or more target subdomains and a mixture of the source domain and preceding target domains;
- and
- recognizing, using the multiple domain discriminators for the target domain, at least one fine-grained object.
A computer system for recognizing fine-grained objects, comprising:
- a processor device operatively coupled to a memory device, the processor device being configured to:
- divide unlabeled training data from a target domain into two or more target subdomains using at least one attribute annotation;
- rank the two or more target subdomains based on a similarity of the at least one attribute annotation to the source domain;
- apply multiple domain discriminators between each of the two or more target subdomains and a mixture of the source domain and preceding target domains;
- and
- recognize at least one fine-grained object using the processing device trained using the multiple domain discriminators for the target domain.
A computer program product for domain adaptation, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computing device to cause the computing device to perform the method comprising:
- dividing, by a processor device, unlabeled training data from a target domain into two or more target subdomains using at least one attribute annotation;
- ranking the two or more target subdomains based on a similarity of the at least one attribute annotation to the source domain;
- applying multiple domain discriminators between each of the two or more target subdomains and a mixture of the source domain and preceding target domains;
- and
- recognizing at least one fine-grained object using the multiple domain discriminators for the target domain.
Dependencies are inferred from the claim text.