Efficient CNN Architecture for Multi-modal Aerial View Object Classification

The NTIRE 2021 workshop features a Multi-modal Aerial View Object Classification Challenge. Its focus is on multi-sensor imagery classification in order to improve the performance of automatic target recognition (ATR) systems. In this paper we describe our entry in this challenge, a method focused on efficiency and low computational time, while maintaining a high level of accuracy. The method is a convolutional neural network with 11 convolutions, 1 max pooling layers and 3 residual blocks which has a total of 373.130 parameters. The method ranks 3rd in the Track 2 (SAR+EO) of the challenge.

[1]  Antonio-Javier Gallego,et al.  Automatic Ship Classification from Optical Aerial Images with Convolutional Neural Networks , 2018, Remote. Sens..

[2]  Takumi Kobayashi,et al.  Large Margin In Softmax Cross-Entropy Loss , 2019, BMVC.

[3]  Chunlei Li,et al.  Livestock classification and counting in quadcopter aerial images using Mask R-CNN , 2020 .

[4]  Hayaru Shouno,et al.  Analysis of function of rectified linear unit used in deep learning , 2015, 2015 International Joint Conference on Neural Networks (IJCNN).

[5]  Yang Song,et al.  Class-Balanced Loss Based on Effective Number of Samples , 2019, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[6]  Tsuyoshi Murata,et al.  Balanced Softmax Cross-Entropy for Incremental Learning , 2021, ICANN.

[7]  Zijun Zhang,et al.  Improved Adam Optimizer for Deep Neural Networks , 2018, 2018 IEEE/ACM 26th International Symposium on Quality of Service (IWQoS).

[8]  Huei-Yung Lin,et al.  Vehicle Detection and Classification in Aerial Images using Convolutional Neural Networks , 2020, VISIGRAPP.

[9]  Nusret Demir,et al.  BUILDING DETECTION FROM SAR IMAGES USING UNET DEEP LEARNING METHOD , 2020 .

[10]  Otmar Loffeld,et al.  Video SAR Moving Target Detection Using Dual Faster R-CNN , 2021, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.

[11]  Hao He,et al.  Air-to-ground multimodal object detection algorithm based on feature association learning , 2019 .

[12]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[13]  Anis Koubaa,et al.  Car Detection using Unmanned Aerial Vehicles: Comparison between Faster R-CNN and YOLOv3 , 2018, 2019 1st International Conference on Unmanned Vehicle Systems-Oman (UVS).

[14]  Geoffrey E. Hinton,et al.  Rectified Linear Units Improve Restricted Boltzmann Machines , 2010, ICML.

[15]  Moustapha Cissé,et al.  Efficient softmax approximation for GPUs , 2016, ICML.

[16]  Filippo Biondi,et al.  Multi-chromatic analysis polarimetric interferometric synthetic aperture radar (MCA-PolInSAR) for urban classification , 2018, International Journal of Remote Sensing.

[17]  Pei-Chun Chen,et al.  Imaging Using Unmanned Aerial Vehicles for Agriculture Land Use Classification , 2020, Agriculture.

[18]  Thomas Oommen,et al.  A comparative analysis of pixel- and object-based detection of landslides from very high-resolution images , 2018, Int. J. Appl. Earth Obs. Geoinformation.

[19]  Ilker Bozcan,et al.  AU-AIR: A Multi-modal Unmanned Aerial Vehicle Dataset for Low Altitude Traffic Surveillance , 2020, 2020 IEEE International Conference on Robotics and Automation (ICRA).

[20]  Jenq-Neng Hwang,et al.  NTIRE 2021 Multi-modal Aerial View Object Classification Challenge , 2021, 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW).

[21]  Allan Aasbjerg Nielsen,et al.  Improving SAR Automatic Target Recognition Models With Transfer Learning From Simulated Data , 2017, IEEE Geoscience and Remote Sensing Letters.