View on GitHub


Realtime adAPtive detecTOR (published at IEEE ICRA 2014)

Download this project as a .zip file Download this project as a tar.gz file

Raptor is an interactive learning approach for object detection models and a showcase how to quickly perform learning and adaptation with large-scale datasets. The details about the method are given in the following paper:

Daniel Göhring, Judy Hoffman, Erik Rodner, Kate Saenko, and Trevor Darrell.
Interactive Adaptation of Real-Time Object Detectors. International Conference on Robotics and Automation (ICRA). 2014



We present a framework for quickly training 2D object detectors for robotic perception. Our method can be used by robotics practitioners to quickly (under 30 seconds per object) build a large-scale real-time perception system. In particular, we show how to create new detectors on the fly using large-scale internet image databases, thus allowing a user to choose among thousands of available categories to build a detection system suitable for the particular robotic application. Furthermore, we show how to adapt these models to the current environment with just a few in-situ images. Experiments on existing 2D benchmarks evaluate the speed, accuracy, and flexibility of our system.



The approach described in the paper is a prototype we build on top of the following methods and software:

Office dataset with bounding boxes

To perform quantitative experiments with object detection and domain adaptation, we used Amazon Mechanical Turk to annotate the Office dataset of Kate Saenko:


  title = {Interactive Adaptation of Real-Time Object Detectors},
  booktitle = {International Conference on Robotics and Automation (ICRA)},
  author = {Daniel Göhring and Judy Hoffman and Erik Rodner and Kate Saenko and Trevor Darrell},
  year = {2014}