Image object detection

Amazon Web Services

Image object detection

Detects instances of real-world entities within an image.

Deploy now
Ready to go in minutes
Ready to go in minutes

Deploy the Image object detection integration to process data on your processing stack in no time. Koyeb integrations are preconfigured applications to ingest, process, and store your data.

Infinite combination
Infinite combination

Compose processing workflows with ready to use integrations and your own code if needed. Build your business logic with our SDKs in Golang, Node.js, and Python or using Docker images.

No infrastructure management
No infrastructure management

Koyeb all-in-one platform is backed by a highly available infrastructure. Build advanced operation pipelines for your data with zero server maintenance.

Integration overview

Image object detection integration uses AWS Rekognition to detect instances of real-world entities within an image (JPEG or PNG). It means that objects like table, vehicle, and wheel; scene like birthday, graduation, and wedding; and concepts like nature, evening, and landscape are detected.

AWS Image Rekognition Object Detection

For each object, scene, and concept, the integration returns one or more labels. Each label provides the object details including:

  • The name of the detected object, scene, or concept.
  • The list of bounding box containing the label and the associated confidence level.
  • The level of confidence that the image includes the label.
  • The parent labels if eligible.

When using the image object detection integration, you can select the AWS region you want to perform the processing.

For ease of use and help you to analyze the result, you can enable the draw bounding boxes option. This option generates a new image file in addition to the JSON response drawing the bounding boxes on the source image.

How it works

Image object detection integration takes an image as input (JPEG or PNG) and returns:

  • A JSON file with the detected labels.
  • A JPEG image drawing the bounding boxes of the object detected on the source image. (If the DrawBoundingBoxes option is enabled)

For instance, if a file name image01.png is uploaded and processed by this integration with the DrawBoundingBoxes option enabled, the result of the processing will be:

  • image01.png.aws-image-label-detection.json
  • image01.png.aws-image-label-detection.jpeg

About the Amazon Web Services Service Provider

Amazon Web Services

Amazon Web Services offers reliable, scalable, and inexpensive cloud computing services.

Discover all Amazon Web Services integrations →
20+

20+

Region

50+

50+

Integrations

60

60

Seconds to deploy

Configure Image object detection

Additional information to get started with the Image object detection integration.

Available settings

  • DrawBoundingBoxes — Generates an image with the bounding boxes returned by the integration.Default value: - Required: No
  • MaxLabels — The maximum number of labels you want the integration to return in the response. The integration returns the specified number of highest confidence labels.Default value: - Required: No
  • MinConfidence — Specifies the minimum confidence level for the labels to return. Amazon Rekognition doesn't return any labels with a confidence lower than this specified value.Default value: - Required: No
  • Region — The AWS region you want to run the processingDefault value: eu-west-1 - Required: Yes

Ready to get started?

Request your invitation now and deploy your processing stack in minutes.

footer-frame