PROJECT TITLE :

Object Detection in Optical Remote Sensing Images Based on Weakly Supervised Learning and High-Level Feature Learning

ABSTRACT:

The abundant spatial and contextual info provided by the advanced remote sensing technology has facilitated subsequent automatic interpretation of the optical remote sensing images (RSIs). In this paper, a novel and effective geospatial object detection framework is proposed by combining the weakly supervised learning (WSL) and high-level feature learning. Initial, deep Boltzmann machine is adopted to infer the spatial and structural info encoded within the low-level and middle-level options to effectively describe objects in optical RSIs. Then, a unique WSL approach is presented to object detection where the coaching sets need only binary labels indicating whether a picture contains the target object or not. Primarily based on the learnt high-level options, it jointly integrates saliency, intraclass compactness, and interclass separability during a Bayesian framework to initialize a collection of coaching examples from weakly labeled pictures and start iterative learning of the object detector. A completely unique analysis criterion is also developed to detect model drift and stop the iterative learning. Comprehensive experiments on three optical RSI data sets have demonstrated the efficacy of the proposed approach in benchmarking with several state-of-the-art supervised-learning-primarily based object detection approaches.