Detect Objects
Detects and locates objects in an image, returning labels, confidence scores, and bounding boxes as JSON. Uses 1 credit per run.
About Detect Objects
Detect Objects finds and locates items in an image, returning each one's label, a confidence score, and a bounding box as structured JSON. Reach for it when you need to count things or know where they are in the frame, not just a written description. It runs on a hosted vision model and costs 1 credit per run.
- Category
- export
- Input
- Accepts: image/png, image/jpeg or image/webp.
- Output
- Outputs: application/json.
- Cost
- Credit-metered
- Memory
- low
Common uses
- Count how many cars or people appear in a surveillance or street photo
- Locate products on a shelf image for a retail audit
- Auto-tag images with the objects they contain for a searchable library
- Feed bounding boxes into a downstream crop or annotation pipeline
- Verify a photo contains an expected item before approving an upload
- Build a quick inventory count from a single warehouse photo
Frequently asked questions
What does the output look like?
A JSON array where each detected object has a label, a confidence score, and a bounding box. That structure plugs directly into code or other JSON-aware tools.
Which image formats are supported?
PNG, JPEG, and WebP.
Does the image get uploaded?
Yes. This Pro tool sends the image to a hosted vision model for detection. It's processed for that run and not retained.
How reliable are the boxes and counts?
Confidence scores let you filter weak detections, but no detector is perfect: overlapping or tiny objects may be missed or merged. Use the scores to set a threshold that fits your task.
What's the cost?
1 credit per image.
Keywords
- detect
- objects
- bounding box
- count
- pro
- vision