Tasks

Zero-Shot Object Detection

Zero-shot object detection is a computer vision task to detect objects and their classes in images, without any prior training or knowledge of the classes. Zero-shot object detection models receive an image as input, as well as a list of candidate classes, and output the bounding boxes and labels where the objects have been detected.

Inputs
Classes

cat, dog, bird

Zero-Shot Object Detection Model
Output

About Zero-Shot Object Detection

Use Cases

Zero-shot object detection models can be used in any object detection application where the detection involves text queries for objects of interest.

Object Search

Zero-shot object detection models can be used in image search. Smartphones, for example, use zero-shot object detection models to detect entities (such as specific places or objects) and allow the user to search for the entity on the internet.

Object Counting

Zero-shot object detection models are used to count instances of objects in a given image. This can include counting the objects in warehouses or stores or the number of visitors in a store. They are also used to manage crowds at events to prevent disasters.

Object Tracking

Zero-shot object detectors can track objects in videos.

Inference

You can infer with zero-shot object detection models through the zero-shot-object-detection pipeline. When calling the pipeline, you just need to specify a path or HTTP link to an image and the candidate labels.

from transformers import pipeline
from PIL import Image

image = Image.open("my-image.png").convert("RGB")

detector = pipeline(model="google/owlvit-base-patch32", task="zero-shot-object-detection")

predictions = detector(
    image,
    candidate_labels=["a photo of a cat", "a photo of a dog"],
)

# [{'score': 0.95,
#   'label': 'a photo of a cat',
#   'box': {'xmin': 180, 'ymin': 71, 'xmax': 271, 'ymax': 178}},
#   ...
# ]

Useful Resources

This page was made possible thanks to the efforts of Victor Guichard

Compatible libraries

Zero-Shot Object Detection demo

No example widget is defined for this task.

Note Contribute by proposing a widget for this task !

Models for Zero-Shot Object Detection
Browse Models (23)
Datasets for Zero-Shot Object Detection
Browse Datasets (0)

No example dataset is defined for this task.

Note Contribute by proposing a dataset for this task !

Spaces using Zero-Shot Object Detection

Note A demo to try the state-of-the-art zero-shot object detection model, OWLv2.

Metrics for Zero-Shot Object Detection
Average Precision
The Average Precision (AP) metric is the Area Under the PR Curve (AUC-PR). It is calculated for each class separately
Mean Average Precision
The Mean Average Precision (mAP) metric is the overall average of the AP values
APα
The APα metric is the Average Precision at the IoU threshold of a α value, for example, AP50 and AP75