-
Notifications
You must be signed in to change notification settings - Fork 33
ClassifyDescriptor
Chaunte W. Lacewell edited this page Aug 26, 2022
·
1 revision
When a new feature vector is extracted from an image, a query to VDMS can be issued to classify that feature vector based on the indexed features in VDMS. VDMS can respond to the query with the label associated with that descriptor.
The bare minimum requirement for this call is the descriptor set we should look to match, the number of closest labels to look for, and descriptor blob of matching dimensions.
- set: name of the set.
- k_closest: How many near neighbors to look at before returning a label
// Classify a descriptor
{
"ClassifyDescriptor": {
"set": "party_faces", // Specify the name of the DescriptorSet
"k_closest": 1 // Classify using one nearest neighbor
}
}
+ blob
// This returns the result as follows:
"ClassifyDescriptors": {
"status": "success",
"label": "Ricky Fort"
}
Visual Data Management System - Intel Labs
FLINNG Library and Performance
Basic Building Blocks
Insert
- AddBlob
- AddBoundingBox
- AddConnection
- AddDescriptor
- AddDescriptorSet
- AddEntity
- AddImage
- AddVideo
- NeoAdd
Query
- ClassifyDescriptor
- FindBlob
- FindBoundingBox
- FindConnection
- FindDescriptor
- FindDescriptorSet
- FindEntity
- FindFrames
- FindImage
- FindVideo
- NeoFind
Update