Towards Better Exploiting Convolutional Neural Networks for Remote Sensing Scene Classification

Abstract

We present an analysis of three possible strategies for exploiting the power of existing convolutional neural networks (ConvNets) in different scenarios from the ones they were trained: full training, fine tuning, and using ConvNets as feature extractors. In many applications, especially including remote sensing, it is not feasible to fully design and train a new ConvNet, as this usually requires a considerable amount of labeled data and demands high computational costs. Therefore, it is important to understand how to obtain the best profit from existing ConvNets. We perform experiments with six popular ConvNets using three remote sensing datasets. We also compare ConvNets in each strategy with existing descriptors and with state-of-the-art baselines. Results point that fine tuning tends to be the best performing strategy. In fact, using the features from the fine-tuned ConvNet with linear SVM obtains the best results. We also achieved state-of-the-art results for the three datasets used.

References in corpus (5)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Caffe: Convolutional Architecture for Fast Feature Embedding
• Going Deeper with Convolutions
• A survey of active learning algorithms for supervised remote sensing image classification
• Transfer Learning from Deep Features for Remote Sensing and Poverty Mapping

Cited by in corpus (52)

• Deep learning in remote sensing: a review
• A trans-disciplinary review of deep learning research for water resources scientists
• Remote Sensing Image Scene Classification Meets Deep Learning: Challenges, Methods, Benchmarks, and Opportunities
• A Comprehensive Survey of Deep Learning in Remote Sensing: Theories, Tools and Challenges for the Community
• Binary Neural Networks: A Survey
• Recent Advances in Convolutional Neural Networks
• Fully Convolutional Networks for Dense Semantic Labelling of High-Resolution Aerial Imagery
• MARTA GANs: Unsupervised Representation Learning for Remote Sensing Image Classification
• Recurrently Exploring Class-wise Attention in A Hybrid Convolutional and Bidirectional LSTM Network for Multi-label Aerial Image Classification
• Multiple instance dense connected convolution neural network for aerial image scene classification
• Hydra: an Ensemble of Convolutional Neural Networks for Geospatial Land Classification
• Dynamic Multi-Context Segmentation of Remote Sensing Images based on Convolutional Networks
• Multi-Granularity Canonical Appearance Pooling for Remote Sensing Scene Classification
• Relation Network for Multi-label Aerial Image Classification
• Comparison of fine-tuning strategies for transfer learning in medical image classification
• A Deep Journey into Super-resolution: A survey
• Exploiting ConvNet Diversity for Flooding Identification
• Multi-Temporal Scene Classification and Scene Change Detection with Correlation based Fusion
• RSI-CB: A Large Scale Remote Sensing Image Classification Benchmark via Crowdsource Data
• MGML: Multi-Granularity Multi-Level Feature Ensemble Network for Remote Sensing Scene Classification
• Panoptic Segmentation Meets Remote Sensing
• All Grains, One Scheme (AGOS): Learning Multi-grain Instance Representation for Aerial Scene Classification
• A framework for remote sensing images processing using deep learning technique
• A CNN-based Spatial Feature Fusion Algorithm for Hyperspectral Imagery Classification
• A global method to identify trees outside of closed-canopy forests with medium-resolution satellite imagery
• In-domain representation learning for remote sensing
• Application of a semantic segmentation convolutional neural network for accurate automatic detection and mapping of solar photovoltaic arrays in aerial imagery
• DML-GANR: Deep Metric Learning With Generative Adversarial Network Regularization for High Spatial Resolution Remote Sensing Image Retrieval
• Deep Learning for Scene Classification: A Survey
• Invariant Deep Compressible Covariance Pooling for Aerial Scene Categorization
• Convolution Neural Network Architecture Learning for Remote Sensing Scene Classification
• Land Use and Land Cover Classification using a Human Group based Particle Swarm Optimization Algorithm with a LSTM classifier on hybrid-pre-processing Remote Sensing Images
• On Creating Benchmark Dataset for Aerial Image Interpretation: Reviews, Guidances and Million-AID
• Window Detection In Facade Imagery: A Deep Learning Approach Using Mask R-CNN
• Cross-Task Transfer for Geotagged Audiovisual Aerial Scene Recognition
• Bio-inspired fine-tuning for selective transfer learning in image classification
• Improving the network traffic classification using the Packet Vision approach
• Learning with less: label-efficient land cover classification at very high spatial resolution using self-supervised deep learning
• Semantic Segmentation of Earth Observation Data Using Multimodal and Multi-scale Deep Networks
• AutoMC: Automated Model Compression based on Domain Knowledge and Progressive search strategy
• On the Adversarial Vulnerabilities of Transfer Learning in Remote Sensing
• Transferable Feature Representation for Visible-to-Infrared Cross-Dataset Human Action Recognition
• MLMA-Net: multi-level multi-attentional learning for multi-label object detection in textile defect images
• Sensing Urban Land-Use Patterns By Integrating Google Tensorflow And Scene-Classification Models
• Enhancement of land-use change modeling using convolutional neural networks and convolutional denoising autoencoders
• Satellite Image Scene Classification via ConvNet with Context Aggregation
• Global Temporal Representation based CNNs for Infrared Action Recognition
• A Comparative Study on Unsupervised Domain Adaptation Approaches for Coffee Crop Mapping
• Deep-learning coupled with novel classification method to classify the urban environment of the developing world
• A Discriminative Learned CNN Embedding for Remote Sensing Image Scene Classification
• A real-time hourly ozone prediction system using deep convolutional neural network
• On the Selective and Invariant Representation of DCNN for High-Resolution Remote Sensing Image Recognition