Discovering social relationships from personal photo collections
View Patent ↗A method of categorizing the relationship of individuals from a collection of photos taken in multiple sets of events includes: searching the photo collection to identify individuals in the photo collection, determining the gender and age range of the identified individuals; and using the identified individuals, their occurrences and co-occurrences, their gender, and their ages to infer social relationships between the individuals based on a set of rules.
1. A method of categorizing the relationship of individuals from a collection of photos taken in multiple sets of events, comprising using a computer for automatically performing the following steps:
a. searching the photo collection to identify individuals in the photo collection, and determining the gender and age range of the identified individuals;
b. determining occurrences of identified individuals and co-occurrences of identified individuals; and
c. performing an inference task using the identified individuals, their occurrences and co-occurrences, their gender, and their age ranges to determine a type of social relationship relationships between a pair of individuals, wherein the types of social relationships include parent, child, spouse, other relative, friend, child's friend, or acquaintance.
2. The method of claim 1 wherein step c wherein the inference task includes using a set of rules including at least one of the following hard rules:
i. parents are older than their children; and
ii. spouses are of opposite gender.
3. The method of claim 2 wherein the set of rules further includes at least one of the following soft rules:
iii. parents and grandparents appear with their children or grandchildren;
iv. friends and relatives are clustered across photos;
v. an acquaintance appears in fewer images than relatives and friends; and
vi. people with closer relationships appear closer in photos than those who do not have a close relationship.
4. The method of claim 1 wherein the inference is performed using a Markov Logic Network (MLN).
5. The method of claim 1 further comprising one or more of the following:
d. adding additional new images to the image collection; or
e. identifying new individuals in the new images.
6. The method of claim 5 further including
f. using the identified new individuals, their occurrences and co-occurrences, their gender, and their ages to infer social relationships between the new individuals and the identified individuals based on the set of rules.
7. A method of categorizing the relationship of individuals from a collection of photos taken in multiple sets of events and belonging to an assumed owner, comprising using a computer for automatically performing the following steps:
a. searching the photo collection to identify individuals in the photo collection, and determining the gender and age range of the identified individuals; and
b. determining occurrences of identified individuals and co-occurrences of identified individuals; and
c. performing an inference task using the identified individuals, their occurrences and co-occurrences, their gender, and their age ranges to determine a type of social relationship between the individuals and the assumed owner of the photo collection, wherein the types of social relationships include parent, child, spouse, other relative, friend, child's friend, or acquaintance.
8. The method of claim 7 wherein step c wherein the inference task includes using a set of rules including at least one of the following hard rules:
i. parents are older than their children; and
ii. spouses are of opposite gender.
9. The method of claim 7 wherein the set of rules further includes at least one of the following soft rules:
iii. children of an assumed owner of the photo collection, occur more frequently in the photo collection;
iv. parents and grandparents appear with their children or grandchildren;
v. friends and relatives are clustered across photos;
vi. an acquaintance appears in fewer images than relatives and friends; and
vii. people with closer relationships appear closer in photos than those who do not have a close relationship.
10. The method of claim 7 wherein the inference is performed using a Markov Logic Network (MLN).
11. The method of claim 7 further comprising one or more of the following:
d. adding additional new images to the image collection; or
e. identifying new individuals in the new images.
12. The method of claim 11 further including
f. using the identified new individuals, their occurrences and co-occurrences, their gender, and their age ranges to infer social relationships between the new individuals and the assumed owner of the photo collection based on the set of rules.