Abstract:In this article, we focus on tackling the problem of weakly supervised object detection from high spatial resolution remote sensing images, which aims to learn detectors with only image-level annotations, i.e., without object location information during the training stage. Although promising results have been achieved, most approaches often fail to provide high-quality initial samples and thus are difficult to obtain optimal object detectors. To address this challenge, a dynamic curriculum learning strategy is proposed to progressively learn the object detectors by feeding training images with increasing difficulty that matches current detection ability. To this end, an entropy-based criterion is firstly designed to evaluate the difficulty for localizing objects in images. Then, an initial curriculum that ranks training images in ascending order of difficulty is generated, in which easy images are selected to provide reliable instances for learning object detectors. With the gained stronger detection ability, the subsequent order in the curriculum for retraining detectors is accordingly adjusted by promoting difficult images as easy ones. In such way, the detectors can be well prepared by training on easy images for learning from more difficult ones and thus gradually improve their detection ability more effectively. Moreover, an effective instance-aware focal loss function for detector learning is developed to alleviate the influence of positive instances of bad quality and meanwhile enhance the discriminative information of class-specific hard negative instances. Comprehensive experiments and comparisons with state-of-the-art methods on two publicly available data sets demonstrate the superiority of our proposed method.

A Combing Top-Down And Bottom-Up Discriminative Dictionaries Learning For Non-Specific Object Detection

A dictionary learning approach for classification: separating the particularity and the commonality

Online Discriminative Dictionary Learning for Image Classification Based on Block-Coordinate Descent Method

A Classification-Oriented Dictionary Learning Model: Explicitly Learning the Particularity and Commonality Across Categories.

Coupled Dictionary Learning for Target Recognition in SAR Images

Object Co-Detection Via Low-Rank and Sparse Representation Dictionary Learning

Discovery of Topical Object in Image Collections

Learning a Multi-class Discriminative Dictionary with Nonredundancy Constraints for Visual Classification

Dictionary Pair Classifier Driven Convolutional Neural Networks for Object Detection

Discriminative Object Tracking Via Sparse Representation and Online Dictionary Learning

Online discriminative dictionary learning for robust object tracking.

Class-Oriented Discriminative Dictionary Learning for Image Classification

Multi-level Discriminative Dictionary Learning with Application to Large Scale Image Classification.

Learning Category-Specific Dictionary and Shared Dictionary for Fine-Grained Image Categorization

Instance-Dictionary Learning for Open-World Object Detection in Autonomous Driving Scenarios

Automatic Weakly Supervised Object Detection from High Spatial Resolution Remote Sensing Images Via Dynamic Curriculum Learning

Jointly Learning Non-negative Projection and Dictionary with Discriminative Graph Constraints for Classification

Learning Group-Based Dictionaries for Discriminative Image Representation.

Learning a Discriminative Dictionary with CNN for Image Classification.

Discriminative Semantic Parts Learning For Object Detection

Learning Discriminative Visual Dictionary for Natural Scene Categorization