Skip to main content

Detect common objects locally and view the annotation boxes and the number of each category.

Select a local image, up to 12 MiB, 4 Megapixels.

The open source model will be loaded from this site at the beginning, and the images will not be uploaded. Model results may be wrong, please review the output and do not use it for safety or medical decisions.

Only COCO's 80 predefined categories are detected, and the results retain the model's English label. The absence of a detection frame does not mean that there are no objects in the picture.

Preview and results

After processing is complete, the results will be displayed here.

Instructions for use

  1. Select a local image and adjust the detection threshold; increasing the threshold will reduce low-confidence detection frames.
  2. Click "Detect and Count" to view the annotated images and the number of each category.
  3. Check for missed detections and repeated detections, and then download annotated images or copy detection data. Unrecognized does not mean there is no object.

Input and output examples

Example input
objects.jpg
Example output
counts / bounding boxes

These are demo inputs and sample results, and do not mean that your data has been processed.

Calculation basis and reference materials

Use COCO-SSD's 80 predefined categories; input up to 4 million pixels, detection map reduced to 1024 px on the longest side. Objects that overlap, are too small, or are not included may be missed, and accurate counting is not guaranteed.

Content check: · About this site and content description

Related tools

Image Joiner and CollageImage AnnotatorLocal picture cutout