Object Detection

Object Detection in Photos

Choose a photo or take one with your phone camera, and AI finds people, cars, animals, food and other everyday objects, draws boxes around them and shows how confident it is. Adjust the confidence threshold, see how many of each kind were found, copy the list and save the image with boxes. Your photo is processed in your browser and never uploaded.

The first time, the AI model (about 33 MB) is downloaded from Hugging Face and then cached by your browser. Your photos are never sent anywhere.

The 80 kinds of objects it can find

The model was trained on the COCO dataset. Things that are not on this list (for example signs with text, buildings, trees or individual products) are not detected, or may be detected as something similar.

  • Person
  • Bicycle
  • Car
  • Motorcycle
  • Airplane
  • Bus
  • Train
  • Truck
  • Boat
  • Traffic light
  • Fire hydrant
  • Stop sign
  • Parking meter
  • Bench
  • Bird
  • Cat
  • Dog
  • Horse
  • Sheep
  • Cow
  • Elephant
  • Bear
  • Zebra
  • Giraffe
  • Backpack
  • Umbrella
  • Handbag
  • Tie
  • Suitcase
  • Frisbee
  • Skis
  • Snowboard
  • Sports ball
  • Kite
  • Baseball bat
  • Baseball glove
  • Skateboard
  • Surfboard
  • Tennis racket
  • Bottle
  • Wine glass
  • Cup
  • Fork
  • Knife
  • Spoon
  • Bowl
  • Banana
  • Apple
  • Sandwich
  • Orange
  • Broccoli
  • Carrot
  • Hot dog
  • Pizza
  • Donut
  • Cake
  • Chair
  • Couch
  • Potted plant
  • Bed
  • Dining table
  • Toilet
  • TV
  • Laptop
  • Mouse
  • Remote
  • Keyboard
  • Cell phone
  • Microwave
  • Oven
  • Toaster
  • Sink
  • Refrigerator
  • Book
  • Clock
  • Vase
  • Scissors
  • Teddy bear
  • Hair dryer
  • Toothbrush
  • Photos are processed only in your browser. They are never uploaded or stored. Only the AI model itself is downloaded from Hugging Face.
  • Detection by AI can miss objects or name them wrongly, especially small, overlapping, dark or unusual ones. Always check the result yourself, and do not rely on it for safety, security or other important decisions.
  • Processing can take from under a second to several seconds depending on your device. A recent browser with WebGPU is fastest.
  • Please respect the privacy of people in your photos, and only share images of others with their permission.
  • Detection model: RF-DETR Medium by Roboflow (Apache-2.0) / Transformers.js (Apache-2.0) / ONNX Runtime Web (MIT).

About this tool

A free object detection tool that finds 80 kinds of everyday things in your photo, such as people, cars, animals, food and furniture, and draws boxes around them with names and confidence scores. Adjust the confidence threshold, count objects by type, copy the list as text and save the image with boxes as a PNG. It also works with photos taken on your phone camera. The AI model (RF-DETR) runs in your browser, so your photos are never uploaded.

How to use

  1. Drop a photo or click the box to choose one. On a smartphone, use “Take a photo (smartphone camera)” to shoot one on the spot.
  2. The first time, the AI model (about 33 MB) is downloaded. Objects are then boxed and labeled with their confidence.
  3. Move the “Confidence threshold” slider to filter the results. Check “Count by type” and tap a type to hide or show its boxes.
  4. Use “Copy the list” to copy the results as text and “Save image with boxes (PNG)” to download the image.

Useful for

  • Roughly counting people, cars or animals in a photo
  • Trying out what AI image recognition can do, for example in a class or workshop
  • Checking what is in blog or product photos to help with tags and descriptions
  • Getting a feel for accuracy before planning a system that uses image recognition

FAQ

What is object detection?

It is an AI technique that finds what is in an image and where, marking each object with a bounding box and a label. Unlike image classification, which gives one label to the whole image, it can find and count several objects at once.

What objects can it detect?

The 80 categories of the COCO dataset, including people, bicycles, cars, dogs, cats, chairs, laptops and pizza (see “The 80 kinds of objects it can find” on the page). It cannot detect buildings, trees, text on signs or specific product names, and it does not identify individual people.

What does the confidence score mean, and what threshold should I use?

It shows how sure the AI is about each detection. Around 50% is a good starting point; lower it if objects are missed and raise it if there are too many wrong boxes.

Are my photos uploaded? Is it free?

Photos are processed only in your browser and never sent to a server. The tool is free with no sign-up; only the AI model (about 33 MB) is downloaded from Hugging Face the first time.

Can I use it to count people?

It works for rough counts, but people who overlap or appear small may be missed. Do not rely on it where exact numbers matter or for safety and security decisions.

Thinking about using AI image recognition in your business?

We build websites and web systems, including tools that use AI in the browser. Feel free to ask us.

Contact us
Contact