VlmClassificationStrategy

Strategy for how zones are sent to the VLM for classification.

from nutrient_sdk import VlmClassificationStrategy

Values

NameValueDescription
VlmClassificationStrategy.ONE_SHOT0Send a single full-page image with bounding box coordinates in the prompt. The VLM performs spatial reasoning to classify each zone by its bbox.
VlmClassificationStrategy.MULTI_ZONE1Crop each zone from the source image and send multiple zone images per VLM call. The VLM classifies each cropped image by its visual content. Supports batching (BatchSize) and parallel execution (MaxConcurrency).

Usage Example 

from nutrient_sdk import VlmClassificationStrategy


# Access enum values
value = VlmClassificationStrategy.ONE_SHOT
print(f"Value: {value}")  # Output: Value: VlmClassificationStrategy.ONE_SHOT
print(f"Integer value: {value.value}")  # Output: Integer value: 0