Time in Zone¶

v3¶

Calculate and track the time spent by tracked objects within one or more defined polygon zones, measure duration of object presence in specific areas (supporting multiple zones where objects are considered 'in zone' if present in any zone), filter detections based on zone membership, reset time tracking when objects leave zones, and enable zone-based analytics, dwell time analysis, and presence monitoring workflows.

How This Block Works¶

This block measures how long each tracked object has been inside one or more defined polygon zones by tracking entry and exit times for each unique track ID. The block supports multiple zones, treating objects as 'in zone' if they are present in any of the defined zones. The block:

1. Receives tracked detection predictions with track IDs, an image with embedded video metadata, and polygon zone definition(s) (single zone or list of zones)
2. Extracts video metadata from the image:
3. Accesses video_metadata from the WorkflowImageData object
4. Extracts fps, frame_number, frame_timestamp, video_identifier, and video source information
5. Uses video_identifier to maintain separate tracking state for different videos
6. Validates that detections have track IDs (tracker_id must be present):
7. Requires detections to come from a tracking block (e.g., Byte Tracker)
8. Each object must have a unique tracker_id that persists across frames
9. Raises an error if tracker_id is missing
10. Normalizes zone input to a list of polygons:
11. Accepts a single polygon zone or a list of polygon zones
12. Automatically wraps single polygons in a list for consistent processing
13. Validates nesting depth and coordinate format for all zones
14. Enables flexible zone input formats (single zone or multiple zones)
15. Initializes or retrieves polygon zones for the video:
16. Creates a list of PolygonZone objects from zone coordinates for each unique zone combination
17. Validates zone coordinates (each zone must be a list of at least 3 points, each with 2 coordinates)
18. Stores zone configurations in an OrderedDict with zone cache (max 100 zone combinations)
19. Uses zone key combining video_identifier and zone coordinates for cache lookup
20. Implements FIFO eviction when cache exceeds 100 zone combinations
21. Configures triggering anchor point (e.g., CENTER, BOTTOM_CENTER) for zone detection
22. Initializes or retrieves time tracking state for the video:
23. Maintains a dictionary tracking when each track_id entered any zone
24. Stores entry timestamps per video using video_identifier
25. Maintains separate tracking state for each video
26. Calculates current timestamp for time measurement:
27. For video files: Calculates timestamp as frame_number / fps
28. For streamed video: Uses frame_timestamp from metadata
29. Provides accurate time measurement for duration calculation
30. Checks which detections are in any zone:
31. Tests each detection against all polygon zones using polygon zone triggers
32. Creates a matrix of zone membership (zones x detections)
33. Uses logical OR operation: objects are considered 'in zone' if they're in ANY of the zones
34. The triggering_anchor determines which point on the bounding box is checked (CENTER, BOTTOM_CENTER, etc.)
35. Returns boolean for each detection indicating zone membership in any zone
36. Updates time tracking for each tracked object:
37. For objects entering any zone: Records entry timestamp if not already tracked
38. For objects in any zone: Calculates time spent as current_timestamp - entry_timestamp
39. For objects leaving all zones:
    • If reset_out_of_zone_detections is True: Removes entry timestamp (resets to 0)
    • If reset_out_of_zone_detections is False: Keeps entry timestamp (continues tracking)
40. Handles out-of-zone detections:
    • If remove_out_of_zone_detections is True: Filters out detections outside all zones from output
    • If remove_out_of_zone_detections is False: Includes out-of-zone detections with time = 0
41. Adds time_in_zone information to each detection:
    • Attaches time_in_zone value (in seconds) to each detection as metadata
    • Objects in any zone: Time represents duration spent in any zone
    • Objects outside all zones: Time is 0 (if not reset) or undefined (if removed)
42. Returns detections with time_in_zone information:
    • Outputs tracked detections enhanced with time_in_zone metadata
    • Filtered or unfiltered based on remove_out_of_zone_detections setting
    • Maintains all original detection properties plus time tracking information

The block maintains persistent tracking state across frames, allowing accurate cumulative time measurement for objects that remain in any zone over multiple frames. Time is measured from when an object first enters any zone (based on its track_id) until the current frame, providing real-time duration tracking. When multiple zones are provided, objects are considered 'in zone' if their anchor point is inside any of the zones, allowing tracking across multiple areas as a single combined zone. The zone cache efficiently manages multiple zone configurations per video using FIFO eviction to limit memory usage. The triggering anchor determines which part of the bounding box is used for zone detection, enabling different zone entry/exit behaviors based on object position.

Common Use Cases¶

• Multi-Zone Dwell Time Analysis: Measure how long objects remain in any of multiple areas for behavior analysis (e.g., measure customer time in any store section, track time spent in multiple parking areas, analyze time in overlapping zones), enabling multi-zone dwell time analytics workflows
• Zone-Based Monitoring: Monitor object presence across multiple defined areas for security and safety (e.g., detect loitering in any restricted area, monitor time in multiple danger zones, track presence across secure zones), enabling multi-zone monitoring workflows
• Retail Analytics: Track customer time across multiple store sections for retail insights (e.g., measure time in any product aisle, analyze shopping patterns across departments, track engagement in multiple zones), enabling multi-zone retail analytics workflows
• Occupancy Management: Measure time objects spend in any of multiple spaces for space utilization (e.g., track vehicle parking duration in multiple lots, measure table occupancy across zones, analyze space usage in multiple areas), enabling multi-zone occupancy management workflows
• Safety Compliance: Monitor time violations across multiple restricted or time-limited zones (e.g., detect extended stays in any hazardous area, monitor time limit violations across zones, track safety compliance in multiple areas), enabling multi-zone safety monitoring workflows
• Traffic Analysis: Measure time vehicles spend in any of multiple traffic zones or intersections (e.g., track time at multiple intersections, measure queue waiting time across zones, analyze traffic flow in multiple areas), enabling multi-zone traffic analytics workflows

Connecting to Other Blocks¶

This block receives an image with embedded video metadata, tracked detections, and zone coordinates (single or multiple zones), and produces timed_detections with time_in_zone metadata:

• After Byte Tracker blocks to measure time for tracked objects across multiple zones (e.g., track time in multiple zones for tracked objects, measure dwell time with consistent IDs across areas, analyze tracked object presence in multiple zones), enabling tracking-to-time workflows
• After zone definition blocks to apply time tracking to multiple defined areas (e.g., measure time across multiple polygon zones, track duration in custom multi-zone configurations, analyze zone-based presence across areas), enabling zone-to-time workflows
• Before logic blocks like Continue If to make decisions based on time in any zone (e.g., continue if time exceeds threshold in any zone, filter based on dwell time across zones, trigger actions on time violations in multiple areas), enabling time-based decision workflows
• Before analysis blocks to analyze time-based metrics across multiple zones (e.g., analyze dwell time patterns across zones, process time-in-zone data for multiple areas, work with duration metrics across zones), enabling time analysis workflows
• Before notification blocks to alert on time violations or thresholds in any zone (e.g., alert on extended stays in any zone, notify on time limit violations across areas, trigger time-based alerts for multiple zones), enabling time-based notification workflows
• Before data storage blocks to record time metrics across multiple zones (e.g., store dwell time data for multiple areas, log time-in-zone metrics across zones, record duration measurements for multiple zones), enabling time metrics logging workflows

Version Differences¶

Enhanced from v2:

• Multiple Zone Support: Supports tracking time across multiple polygon zones simultaneously, where objects are considered 'in zone' if they're present in any of the defined zones, enabling multi-zone time tracking and analysis
• Flexible Zone Input: Accepts either a single polygon zone or a list of polygon zones, automatically normalizing the input to handle both formats seamlessly
• Zone Cache Management: Implements a zone cache with FIFO eviction (max 100 zone combinations) to efficiently manage multiple zone configurations per video while limiting memory usage
• Combined Zone Logic: Uses logical OR operation across all zones, allowing tracking across multiple areas as a unified zone system for comprehensive presence monitoring
• Enhanced Zone Key System: Uses combined zone keys (video_identifier + zone coordinates) for cache lookup, enabling efficient storage and retrieval of zone configurations

Requirements¶

This block requires tracked detections with tracker_id information (detections must come from a tracking block like Byte Tracker). The zone can be a single polygon or a list of polygons, where each polygon must be defined as a list of at least 3 points, with each point being a list or tuple of exactly 2 coordinates (x, y). The image's video_metadata should include frame rate (fps) for video files or frame timestamps for streamed video to calculate accurate time measurements. The block maintains persistent tracking state across frames for each video, so it should be used in video workflows where frames are processed sequentially. For accurate time measurement, detections should be provided consistently across frames with valid track IDs. When multiple zones are provided, objects are considered 'in zone' if they're present in any of the zones.

Type identifier¶

Use the following identifier in step "type" field: roboflow_core/time_in_zone@v3to add the block as as step in your workflow.

Properties¶

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

Runtime compatibility¶

soft — runtime hosted_serverless, dedicated_deployment; execution remote; input video

Block keeps per-video state in process memory (keyed by video_metadata.video_identifier). With remote step execution on stateless or multi-replica HTTP runtimes, successive requests may be served by different worker processes, so the state resets between calls and the output is meaningless for tracking / counting / aggregation. Use local step execution in an InferencePipeline for stable cross-frame results.

soft — input image

Block depends on temporal context from video or repeated-frame workflows. With a still image/photo, there is no meaningful history to track, compare, aggregate, or visualize, so the block provides little or no benefit.

Available Connections¶

Input and Output Bindings¶

The available connections depend on its binding kinds. Check what binding kinds Time in Zone in version v3 has.

{
    "name": "<your_step_name_here>",
    "type": "roboflow_core/time_in_zone@v3",
    "image": "$inputs.image",
    "detections": "$steps.object_detection_model.predictions",
    "zone": [
        [
            100,
            100
        ],
        [
            100,
            200
        ],
        [
            300,
            200
        ],
        [
            300,
            100
        ]
    ],
    "triggering_anchor": "CENTER",
    "remove_out_of_zone_detections": true,
    "reset_out_of_zone_detections": true
}

v2¶

Calculate and track the time spent by tracked objects within a defined polygon zone, measure duration of object presence in specific areas, filter detections based on zone membership, reset time tracking when objects leave zones, and enable zone-based analytics, dwell time analysis, and presence monitoring workflows.

How This Block Works¶

This block measures how long each tracked object has been inside a defined polygon zone by tracking entry and exit times for each unique track ID. The block:

1. Receives tracked detection predictions with track IDs, an image with embedded video metadata, and a polygon zone definition
2. Extracts video metadata from the image:
3. Accesses video_metadata from the WorkflowImageData object
4. Extracts fps, frame_number, frame_timestamp, video_identifier, and video source information
5. Uses video_identifier to maintain separate tracking state for different videos
6. Validates that detections have track IDs (tracker_id must be present):
7. Requires detections to come from a tracking block (e.g., Byte Tracker)
8. Each object must have a unique tracker_id that persists across frames
9. Raises an error if tracker_id is missing
10. Initializes or retrieves a polygon zone for the video:
11. Creates a PolygonZone object from zone coordinates for each unique video
12. Validates zone coordinates (must be a list of at least 3 points, each with 2 coordinates)
13. Stores zone configuration per video using video_identifier
14. Configures triggering anchor point (e.g., CENTER, BOTTOM_CENTER) for zone detection
15. Initializes or retrieves time tracking state for the video:
16. Maintains a dictionary tracking when each track_id entered the zone
17. Stores entry timestamps per video using video_identifier
18. Maintains separate tracking state for each video
19. Calculates current timestamp for time measurement:
20. For video files: Calculates timestamp as frame_number / fps
21. For streamed video: Uses frame_timestamp from metadata
22. Provides accurate time measurement for duration calculation
23. Checks which detections are in the zone:
24. Uses polygon zone trigger to test if each detection's anchor point is inside the zone
25. The triggering_anchor determines which point on the bounding box is checked (CENTER, BOTTOM_CENTER, etc.)
26. Returns boolean for each detection indicating zone membership
27. Updates time tracking for each tracked object:
28. For objects entering the zone: Records entry timestamp if not already tracked
29. For objects in the zone: Calculates time spent as current_timestamp - entry_timestamp
30. For objects leaving the zone:
    • If reset_out_of_zone_detections is True: Removes entry timestamp (resets to 0)
    • If reset_out_of_zone_detections is False: Keeps entry timestamp (continues tracking)
31. Handles out-of-zone detections:
32. If remove_out_of_zone_detections is True: Filters out detections outside the zone from output
33. If remove_out_of_zone_detections is False: Includes out-of-zone detections with time = 0
34. Adds time_in_zone information to each detection:
    • Attaches time_in_zone value (in seconds) to each detection as metadata
    • Objects in zone: Time represents duration spent in zone
    • Objects outside zone: Time is 0 (if not reset) or undefined (if removed)
35. Returns detections with time_in_zone information:
    • Outputs tracked detections enhanced with time_in_zone metadata
    • Filtered or unfiltered based on remove_out_of_zone_detections setting
    • Maintains all original detection properties plus time tracking information

The block maintains persistent tracking state across frames, allowing accurate cumulative time measurement for objects that remain in the zone over multiple frames. Time is measured from when an object first enters the zone (based on its track_id) until the current frame, providing real-time duration tracking. The zone is defined as a polygon with multiple points, allowing flexible area definitions. The triggering anchor determines which part of the bounding box is used for zone detection, enabling different zone entry/exit behaviors based on object position.

Common Use Cases¶

• Dwell Time Analysis: Measure how long objects remain in specific areas for behavior analysis (e.g., measure customer dwell time in store sections, track time spent in parking spaces, analyze time in waiting areas), enabling dwell time analytics workflows
• Zone-Based Monitoring: Monitor object presence in defined areas for security and safety (e.g., detect loitering in restricted areas, monitor time in danger zones, track presence in secure zones), enabling zone monitoring workflows
• Retail Analytics: Track customer time in different store sections for retail insights (e.g., measure time in product aisles, analyze shopping patterns, track department engagement), enabling retail analytics workflows
• Occupancy Management: Measure time objects spend in spaces for space utilization (e.g., track vehicle parking duration, measure table occupancy time, analyze space usage patterns), enabling occupancy management workflows
• Safety Compliance: Monitor time violations in restricted or time-limited zones (e.g., detect extended stays in hazardous areas, monitor time limit violations, track safety compliance), enabling safety monitoring workflows
• Traffic Analysis: Measure time vehicles spend in traffic zones or intersections (e.g., track time at intersections, measure queue waiting time, analyze traffic flow patterns), enabling traffic analytics workflows

Connecting to Other Blocks¶

This block receives an image with embedded video metadata, tracked detections, and zone coordinates, and produces timed_detections with time_in_zone metadata:

• After Byte Tracker blocks to measure time for tracked objects (e.g., track time in zones for tracked objects, measure dwell time with consistent IDs, analyze tracked object presence), enabling tracking-to-time workflows
• After zone definition blocks to apply time tracking to defined areas (e.g., measure time in polygon zones, track duration in custom zones, analyze zone-based presence), enabling zone-to-time workflows
• Before logic blocks like Continue If to make decisions based on time in zone (e.g., continue if time exceeds threshold, filter based on dwell time, trigger actions on time violations), enabling time-based decision workflows
• Before analysis blocks to analyze time-based metrics (e.g., analyze dwell time patterns, process time-in-zone data, work with duration metrics), enabling time analysis workflows
• Before notification blocks to alert on time violations or thresholds (e.g., alert on extended stays, notify on time limit violations, trigger time-based alerts), enabling time-based notification workflows
• Before data storage blocks to record time metrics (e.g., store dwell time data, log time-in-zone metrics, record duration measurements), enabling time metrics logging workflows

Version Differences¶

Enhanced from v1:

• Simplified Input: Uses image input that contains embedded video metadata instead of requiring a separate metadata field, simplifying workflow connections and reducing input complexity
• Improved Integration: Better integration with image-based workflows since video metadata is accessed directly from the image object rather than requiring separate metadata input
• Streamlined Workflow: Reduces the number of inputs needed, making it easier to connect in workflows where image and metadata come from the same source

Requirements¶

This block requires tracked detections with tracker_id information (detections must come from a tracking block like Byte Tracker). The zone must be defined as a list of at least 3 points, where each point is a list or tuple of exactly 2 coordinates (x, y). The image's video_metadata should include frame rate (fps) for video files or frame timestamps for streamed video to calculate accurate time measurements. The block maintains persistent tracking state across frames for each video, so it should be used in video workflows where frames are processed sequentially. For accurate time measurement, detections should be provided consistently across frames with valid track IDs.

Type identifier¶

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

Properties¶

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

Runtime compatibility¶

soft — runtime hosted_serverless, dedicated_deployment; execution remote; input video

Block keeps per-video state in process memory (keyed by video_metadata.video_identifier). With remote step execution on stateless or multi-replica HTTP runtimes, successive requests may be served by different worker processes, so the state resets between calls and the output is meaningless for tracking / counting / aggregation. Use local step execution in an InferencePipeline for stable cross-frame results.

soft — input image

Block depends on temporal context from video or repeated-frame workflows. With a still image/photo, there is no meaningful history to track, compare, aggregate, or visualize, so the block provides little or no benefit.

Available Connections¶

Input and Output Bindings¶

The available connections depend on its binding kinds. Check what binding kinds Time in Zone in version v2 has.

{
    "name": "<your_step_name_here>",
    "type": "roboflow_core/time_in_zone@v2",
    "image": "$inputs.image",
    "detections": "$steps.object_detection_model.predictions",
    "zone": [
        [
            100,
            100
        ],
        [
            100,
            200
        ],
        [
            300,
            200
        ],
        [
            300,
            100
        ]
    ],
    "triggering_anchor": "CENTER",
    "remove_out_of_zone_detections": true,
    "reset_out_of_zone_detections": true
}

v1¶

Calculate and track the time spent by tracked objects within a defined polygon zone, measure duration of object presence in specific areas, filter detections based on zone membership, reset time tracking when objects leave zones, and enable zone-based analytics, dwell time analysis, and presence monitoring workflows.

How This Block Works¶

This block measures how long each tracked object has been inside a defined polygon zone by tracking entry and exit times for each unique track ID. The block:

1. Receives tracked detection predictions with track IDs, an image, video metadata, and a polygon zone definition
2. Validates that detections have track IDs (tracker_id must be present):
3. Requires detections to come from a tracking block (e.g., Byte Tracker)
4. Each object must have a unique tracker_id that persists across frames
5. Raises an error if tracker_id is missing
6. Initializes or retrieves a polygon zone for the video:
7. Creates a PolygonZone object from zone coordinates for each unique video
8. Validates zone coordinates (must be a list of at least 3 points, each with 2 coordinates)
9. Stores zone configuration per video using video_identifier
10. Configures triggering anchor point (e.g., CENTER, BOTTOM_CENTER) for zone detection
11. Initializes or retrieves time tracking state for the video:
12. Maintains a dictionary tracking when each track_id entered the zone
13. Stores entry timestamps per video using video_identifier
14. Maintains separate tracking state for each video
15. Calculates current timestamp for time measurement:
16. For video files: Calculates timestamp as frame_number / fps
17. For streamed video: Uses frame_timestamp from metadata
18. Provides accurate time measurement for duration calculation
19. Checks which detections are in the zone:
20. Uses polygon zone trigger to test if each detection's anchor point is inside the zone
21. The triggering_anchor determines which point on the bounding box is checked (CENTER, BOTTOM_CENTER, etc.)
22. Returns boolean for each detection indicating zone membership
23. Updates time tracking for each tracked object:
24. For objects entering the zone: Records entry timestamp if not already tracked
25. For objects in the zone: Calculates time spent as current_timestamp - entry_timestamp
26. For objects leaving the zone:
    • If reset_out_of_zone_detections is True: Removes entry timestamp (resets to 0)
    • If reset_out_of_zone_detections is False: Keeps entry timestamp (continues tracking)
27. Handles out-of-zone detections:
28. If remove_out_of_zone_detections is True: Filters out detections outside the zone from output
29. If remove_out_of_zone_detections is False: Includes out-of-zone detections with time = 0
30. Adds time_in_zone information to each detection:
31. Attaches time_in_zone value (in seconds) to each detection as metadata
32. Objects in zone: Time represents duration spent in zone
33. Objects outside zone: Time is 0 (if not reset) or undefined (if removed)
34. Returns detections with time_in_zone information:
    • Outputs tracked detections enhanced with time_in_zone metadata
    • Filtered or unfiltered based on remove_out_of_zone_detections setting
    • Maintains all original detection properties plus time tracking information

The block maintains persistent tracking state across frames, allowing accurate cumulative time measurement for objects that remain in the zone over multiple frames. Time is measured from when an object first enters the zone (based on its track_id) until the current frame, providing real-time duration tracking. The zone is defined as a polygon with multiple points, allowing flexible area definitions. The triggering anchor determines which part of the bounding box is used for zone detection, enabling different zone entry/exit behaviors based on object position.

Common Use Cases¶

• Dwell Time Analysis: Measure how long objects remain in specific areas for behavior analysis (e.g., measure customer dwell time in store sections, track time spent in parking spaces, analyze time in waiting areas), enabling dwell time analytics workflows
• Zone-Based Monitoring: Monitor object presence in defined areas for security and safety (e.g., detect loitering in restricted areas, monitor time in danger zones, track presence in secure zones), enabling zone monitoring workflows
• Retail Analytics: Track customer time in different store sections for retail insights (e.g., measure time in product aisles, analyze shopping patterns, track department engagement), enabling retail analytics workflows
• Occupancy Management: Measure time objects spend in spaces for space utilization (e.g., track vehicle parking duration, measure table occupancy time, analyze space usage patterns), enabling occupancy management workflows
• Safety Compliance: Monitor time violations in restricted or time-limited zones (e.g., detect extended stays in hazardous areas, monitor time limit violations, track safety compliance), enabling safety monitoring workflows
• Traffic Analysis: Measure time vehicles spend in traffic zones or intersections (e.g., track time at intersections, measure queue waiting time, analyze traffic flow patterns), enabling traffic analytics workflows

Connecting to Other Blocks¶

This block receives tracked detections, image, video metadata, and zone coordinates, and produces timed_detections with time_in_zone metadata:

• After Byte Tracker blocks to measure time for tracked objects (e.g., track time in zones for tracked objects, measure dwell time with consistent IDs, analyze tracked object presence), enabling tracking-to-time workflows
• After zone definition blocks to apply time tracking to defined areas (e.g., measure time in polygon zones, track duration in custom zones, analyze zone-based presence), enabling zone-to-time workflows
• Before logic blocks like Continue If to make decisions based on time in zone (e.g., continue if time exceeds threshold, filter based on dwell time, trigger actions on time violations), enabling time-based decision workflows
• Before analysis blocks to analyze time-based metrics (e.g., analyze dwell time patterns, process time-in-zone data, work with duration metrics), enabling time analysis workflows
• Before notification blocks to alert on time violations or thresholds (e.g., alert on extended stays, notify on time limit violations, trigger time-based alerts), enabling time-based notification workflows
• Before data storage blocks to record time metrics (e.g., store dwell time data, log time-in-zone metrics, record duration measurements), enabling time metrics logging workflows

Requirements¶

This block requires tracked detections with tracker_id information (detections must come from a tracking block like Byte Tracker). The zone must be defined as a list of at least 3 points, where each point is a list or tuple of exactly 2 coordinates (x, y). The block requires video metadata with frame rate (fps) for video files or frame timestamps for streamed video to calculate accurate time measurements. The block maintains persistent tracking state across frames for each video, so it should be used in video workflows where frames are processed sequentially. For accurate time measurement, detections should be provided consistently across frames with valid track IDs.

Type identifier¶

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

Properties¶

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

Runtime compatibility¶

soft — runtime hosted_serverless, dedicated_deployment; execution remote; input video

Block keeps per-video state in process memory (keyed by video_metadata.video_identifier). With remote step execution on stateless or multi-replica HTTP runtimes, successive requests may be served by different worker processes, so the state resets between calls and the output is meaningless for tracking / counting / aggregation. Use local step execution in an InferencePipeline for stable cross-frame results.

soft — input image

Block depends on temporal context from video or repeated-frame workflows. With a still image/photo, there is no meaningful history to track, compare, aggregate, or visualize, so the block provides little or no benefit.

Available Connections¶

Input and Output Bindings¶

The available connections depend on its binding kinds. Check what binding kinds Time in Zone in version v1 has.

{
    "name": "<your_step_name_here>",
    "type": "roboflow_core/time_in_zone@v1",
    "image": "$inputs.image",
    "metadata": "<block_does_not_provide_example>",
    "detections": "$steps.object_detection_model.predictions",
    "zone": "$inputs.zones",
    "triggering_anchor": "CENTER",
    "remove_out_of_zone_detections": true,
    "reset_out_of_zone_detections": true
}