Allen Institute for AI

Detection

Object detection is the task of identifying and locating objects in an image. Object detection models are typically trained and evaluated on the MS-COCO dataset that has 80 object categories.

Try it for yourself
1. Upload an Image (or choose one from the examples)
Examples...
Image:
Click to upload your own image
2. Run a model

YoloV3

YOLOv3: An Incremental ImprovementJoseph RedmonAli FarhadiarXiv2018

As the name “You Only Look Once” suggests, this is a standard single shot model. Only examining the features of the image once allows this model to detect objects incredibly quickly. While this comes at a small cost to accuracy, Yolo is nearly comparable in terms of accuracy to the larger two stage detectors.

Faster R-CNN

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal NetworksShaoqing RenKaiming HeRoss B. GirshickJian SunNeurIPS2015

This model is an example of a two stage detector, meaning it first selects regions which are likely to contain objects and then locates the objects within these regions. However, unlike some of its predecessors, Faster R-CNN is able to maintain a reasonable speed by sharing image features between its two stages. While much slower than single shot detector like yolo, it has a slightly higher accuracy.