Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Videos

Training-free, generic object detection using locally adaptive regression kernels.

Hae Jong Seo¹, Peyman Milanfar

¹University of California, Santa Cruz, 1156 High Street, Mailcode SOE2, Santa Cruz, CA 95064, USA. rokaf@soe.ucsc.edu

IEEE Transactions on Pattern Analysis and Machine Intelligence

|July 17, 2010

Summary

This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Denoising: a powerful building block for imaging, inverse problems and machine learning.

Philosophical transactions. Series A, Mathematical, physical, and engineering sciences·2025

Same author

DVMark: A Deep Multiscale Framework for Video Watermarking.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2023

Same author

Mobile Computational Photography: A Tour.

Annual review of vision science·2021

Same author

Better Compression With Deep Pre-Editing.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2021

Same author

Deep K-SVD Denoising.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2021

Same author

Local Kernels that Approximate Bayesian Regularization and Proximal Operators.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2019

This study introduces a novel visual object detection algorithm that finds similar objects using just one example. It requires no prior training or image preprocessing, enabling efficient and versatile object localization.

Area of Science:

Computer Vision
Machine Learning
Image Analysis

Background:

Traditional object detection often requires extensive training datasets and prior knowledge of object characteristics.
Preprocessing and segmentation steps can be computationally intensive and limit algorithm applicability.
A need exists for generic, training-free algorithms capable of detecting objects from a single example.

Purpose of the Study:

To develop a generic detection and localization algorithm for visual objects without requiring prior training.
To enable object searching using only a single example of the object of interest.
To create a method that operates without preprocessing or segmentation of the target image.

Main Methods:

Computation of local regression kernels as descriptors from a query image.

Related Experiment Videos

Extraction of salient features from descriptors and comparison using a matrix generalization of cosine similarity.

Generation of a scalar resemblance map indicating pixel-wise similarity likelihood.

Application of nonparametric significance tests and nonmaxima suppression for detection and localization.

Extension of the approach to handle variations in scale and rotation.

Main Results:

The algorithm successfully detects and localizes objects similar to a single query example.
Demonstrated high performance on challenging datasets across diverse contexts and imaging conditions.
The method is robust to significant variations in object scale and rotation.
Optimality properties of the algorithm are supported by a naive-Bayes framework.

Conclusions:

The proposed algorithm offers a powerful, training-free approach for visual object detection and localization.
Its ability to work with a single example and without preprocessing makes it highly versatile.
The method shows significant promise for applications requiring rapid and adaptable object recognition.