VLM as Detector

v2

Class: VLMAsDetectorBlockV2 (there are multiple versions of this block)

Source: inference.core.workflows.core_steps.formatters.vlm_as_detector.v2.VLMAsDetectorBlockV2

Warning: This block has multiple versions. Please refer to the specific version for details. You can learn more about how versions work here: Versioning

The block expects string input that would be produced by blocks exposing Large Language Models (LLMs) and Visual Language Models (VLMs). Input is parsed to object-detection prediction and returned as block output.

Accepted formats:

• valid JSON strings

• JSON documents wrapped with Markdown tags

Example

{"my": "json"}

Details regarding block behavior:

• error_status is set True whenever parsing cannot be completed

• in case of multiple markdown blocks with raw JSON content - only first will be parsed

Type identifier

Use the following identifier in step "type" field: roboflow_core/vlm_as_detector@v2to add the block as as step in your workflow.

Properties

Name	Type	Description	Refs
name	str	Enter a unique identifier for this step..	❌
classes	List[str]	List of all classes used by the model, required to generate mapping between class name and class id..	✅
model_type	str	Type of the model that generated prediction.	❌
task_type	str	Task type to performed by model..	❌

The Refs column marks possibility to parametrise the property with dynamic values available in workflow runtime. See Bindings for more info.

Available Connections

Compatible Blocks

Check what blocks you can connect to VLM as Detector in version v2.

• inputs: Image Preprocessing, Relative Static Crop, Background Color Visualization, Line Counter Visualization, SIFT, Dynamic Zone, OpenAI, Reference Path Visualization, Grid Visualization, Blur Visualization, Image Contours, Pixelate Visualization, Llama 3.2 Vision, Camera Focus, Color Visualization, Ellipse Visualization, Label Visualization, Image Slicer, Crop Visualization, Stitch Images, Polygon Zone Visualization, Dot Visualization, Bounding Box Visualization, Florence-2 Model, Image Slicer, Perspective Correction, Size Measurement, Dimension Collapse, Image Blur, Stability AI Image Generation, Dynamic Crop, Image Convert Grayscale, Stability AI Inpainting, Clip Comparison, Clip Comparison, Model Comparison Visualization, Triangle Visualization, Classification Label Visualization, Trace Visualization, Anthropic Claude, Buffer, Depth Estimation, Corner Visualization, SIFT Comparison, Camera Calibration, Absolute Static Crop, Keypoint Visualization, Mask Visualization, Image Threshold, Florence-2 Model, Halo Visualization, Circle Visualization, Google Gemini, Polygon Visualization
• outputs: Line Counter Visualization, Background Color Visualization, Roboflow Dataset Upload, Gaze Detection, Detections Stabilizer, Detections Transformation, Object Detection Model, Roboflow Dataset Upload, Instance Segmentation Model, Color Visualization, Label Visualization, Detections Filter, Dot Visualization, Overlap Filter, Byte Tracker, Object Detection Model, Bounding Box Visualization, Detections Consensus, Template Matching, Single-Label Classification Model, Path Deviation, Dynamic Crop, Model Comparison Visualization, Corner Visualization, SIFT Comparison, Multi-Label Classification Model, Byte Tracker, Line Counter, Detection Offset, Keypoint Visualization, Mask Visualization, Florence-2 Model, Halo Visualization, Polygon Zone Visualization, Polygon Visualization, Instance Segmentation Model, Twilio SMS Notification, Dynamic Zone, Roboflow Custom Metadata, Reference Path Visualization, Blur Visualization, Detections Classes Replacement, Pixelate Visualization, Keypoint Detection Model, Detections Stitch, Ellipse Visualization, Byte Tracker, Crop Visualization, Line Counter, Florence-2 Model, Stitch OCR Detections, Webhook Sink, Email Notification, Perspective Correction, Multi-Label Classification Model, Size Measurement, Velocity, Time in Zone, Model Monitoring Inference Aggregator, Triangle Visualization, Trace Visualization, Classification Label Visualization, Keypoint Detection Model, Single-Label Classification Model, Distance Measurement, Segment Anything 2 Model, Path Deviation, Time in Zone, Circle Visualization, Slack Notification, Detections Merge

Input and Output Bindings

The available connections depend on its binding kinds. Check what binding kinds VLM as Detector in version v2 has.

Bindings

• input

  • image (image): The image which was the base to generate VLM prediction.
  • vlm_output (language_model_output): The string with raw classification prediction to parse..
  • classes (list_of_values): List of all classes used by the model, required to generate mapping between class name and class id..

• output

  • error_status (boolean): Boolean flag.
  • predictions (object_detection_prediction): Prediction with detected bounding boxes in form of sv.Detections(...) object.
  • inference_id (inference_id): Inference identifier.

Example JSON definition of step VLM as Detector in version v2

{
    "name": "<your_step_name_here>",
    "type": "roboflow_core/vlm_as_detector@v2",
    "image": "$inputs.image",
    "vlm_output": [
        "$steps.lmm.output"
    ],
    "classes": [
        "$steps.lmm.classes",
        "$inputs.classes",
        [
            "class_a",
            "class_b"
        ]
    ],
    "model_type": [
        "google-gemini",
        "anthropic-claude",
        "florence-2"
    ],
    "task_type": "<block_does_not_provide_example>"
}

v1

Class: VLMAsDetectorBlockV1 (there are multiple versions of this block)

Source: inference.core.workflows.core_steps.formatters.vlm_as_detector.v1.VLMAsDetectorBlockV1

Warning: This block has multiple versions. Please refer to the specific version for details. You can learn more about how versions work here: Versioning

The block expects string input that would be produced by blocks exposing Large Language Models (LLMs) and Visual Language Models (VLMs). Input is parsed to object-detection prediction and returned as block output.

Accepted formats:

• valid JSON strings

• JSON documents wrapped with Markdown tags

Example

{"my": "json"}

Details regarding block behavior:

• error_status is set True whenever parsing cannot be completed

• in case of multiple markdown blocks with raw JSON content - only first will be parsed

Type identifier

Use the following identifier in step "type" field: roboflow_core/vlm_as_detector@v1to add the block as as step in your workflow.

Properties

Name	Type	Description	Refs
name	str	Enter a unique identifier for this step..	❌
classes	List[str]	List of all classes used by the model, required to generate mapping between class name and class id..	✅
model_type	str	Type of the model that generated prediction.	❌
task_type	str	Task type to performed by model..	❌

The Refs column marks possibility to parametrise the property with dynamic values available in workflow runtime. See Bindings for more info.

Available Connections

Compatible Blocks

Check what blocks you can connect to VLM as Detector in version v1.

• inputs: Image Preprocessing, Relative Static Crop, Background Color Visualization, Line Counter Visualization, SIFT, Dynamic Zone, OpenAI, Reference Path Visualization, Grid Visualization, Blur Visualization, Image Contours, Pixelate Visualization, Llama 3.2 Vision, Camera Focus, Color Visualization, Ellipse Visualization, Label Visualization, Image Slicer, Crop Visualization, Stitch Images, Polygon Zone Visualization, Dot Visualization, Bounding Box Visualization, Florence-2 Model, Image Slicer, Perspective Correction, Size Measurement, Dimension Collapse, Image Blur, Stability AI Image Generation, Dynamic Crop, Image Convert Grayscale, Stability AI Inpainting, Clip Comparison, Clip Comparison, Model Comparison Visualization, Triangle Visualization, Classification Label Visualization, Trace Visualization, Anthropic Claude, Buffer, Depth Estimation, Corner Visualization, SIFT Comparison, Camera Calibration, Absolute Static Crop, Keypoint Visualization, Mask Visualization, Image Threshold, Florence-2 Model, Halo Visualization, Circle Visualization, Google Gemini, Polygon Visualization
• outputs: Line Counter Visualization, Background Color Visualization, OpenAI, Roboflow Dataset Upload, Gaze Detection, Detections Stabilizer, Detections Transformation, Object Detection Model, LMM For Classification, Cache Set, Llama 3.2 Vision, Roboflow Dataset Upload, Instance Segmentation Model, Color Visualization, Pixel Color Count, Label Visualization, YOLO-World Model, Detections Filter, Dot Visualization, Overlap Filter, Byte Tracker, Object Detection Model, Bounding Box Visualization, Detections Consensus, Template Matching, Single-Label Classification Model, Google Vision OCR, Path Deviation, Dynamic Crop, Clip Comparison, Model Comparison Visualization, Anthropic Claude, Corner Visualization, SIFT Comparison, Multi-Label Classification Model, Byte Tracker, Line Counter, Detection Offset, Keypoint Visualization, Mask Visualization, Image Threshold, Florence-2 Model, Halo Visualization, Polygon Zone Visualization, Google Gemini, Polygon Visualization, Image Preprocessing, Instance Segmentation Model, Twilio SMS Notification, Dynamic Zone, Roboflow Custom Metadata, Reference Path Visualization, CLIP Embedding Model, Blur Visualization, Detections Classes Replacement, Pixelate Visualization, Keypoint Detection Model, Detections Stitch, OpenAI, Ellipse Visualization, Byte Tracker, Crop Visualization, Line Counter, Florence-2 Model, Stitch OCR Detections, CogVLM, LMM, Webhook Sink, Cache Get, Email Notification, Perspective Correction, Multi-Label Classification Model, Size Measurement, Velocity, Time in Zone, Stability AI Image Generation, Image Blur, Stability AI Inpainting, Model Monitoring Inference Aggregator, Triangle Visualization, Trace Visualization, Classification Label Visualization, Keypoint Detection Model, Single-Label Classification Model, Distance Measurement, Segment Anything 2 Model, Path Deviation, Time in Zone, Circle Visualization, Slack Notification, Detections Merge, Local File Sink

Input and Output Bindings

The available connections depend on its binding kinds. Check what binding kinds VLM as Detector in version v1 has.

Bindings

• input

  • image (image): The image which was the base to generate VLM prediction.
  • vlm_output (language_model_output): The string with raw classification prediction to parse..
  • classes (list_of_values): List of all classes used by the model, required to generate mapping between class name and class id..

• output

  • error_status (boolean): Boolean flag.
  • predictions (object_detection_prediction): Prediction with detected bounding boxes in form of sv.Detections(...) object.
  • inference_id (string): String value.

Example JSON definition of step VLM as Detector in version v1

{
    "name": "<your_step_name_here>",
    "type": "roboflow_core/vlm_as_detector@v1",
    "image": "$inputs.image",
    "vlm_output": [
        "$steps.lmm.output"
    ],
    "classes": [
        "$steps.lmm.classes",
        "$inputs.classes",
        [
            "class_a",
            "class_b"
        ]
    ],
    "model_type": [
        "google-gemini",
        "anthropic-claude",
        "florence-2"
    ],
    "task_type": "<block_does_not_provide_example>"
}