| Presentation | 2017-08-24 Deep Learning for Target Classification from SAR Imagery Hidetoshi Furukawa, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | This report deals with translation invariance of convolutional neural networks (CNNs) for automatic target recognition (ATR) from synthetic aperture radar (SAR) imagery. In particular, the translation invariance of CNNs for SAR ATR represents the robustness against misalignment of target chips extracted from SAR images. To understand the translation invariance of the CNNs, we trained CNNs which classify the target chips from the MSTAR into the ten classes under the condition of with and without data augmentation, and then visualized the translation invariance of the CNNs. According to our results, even if we use a deep residual network, the translation invariance of the CNN without data augmentation using the aligned images such as the MSTAR target chips is not so large. A more important factor of translation invariance is the use of augmented training data. Furthermore, our CNN using augmented training data achieved a state-of-the-art classification accuracy of 99.6%. These results show an importance of domain-specific data augmentation. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | synthetic aperture radar (SAR) / automatic target recognition (ATR) / target classification / deep learning / convolutional neural network (CNN) / data augmentation / translation invariance / residual network |
| Paper # | SANE2017-30 |
| Date of Issue | 2017-08-17 (SANE) |
| Conference Information | |
| Committee | SANE |
|---|---|
| Conference Date | 2017/8/24(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | OIT UMEDA Campus |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Navigation, Trafic control, Radar, Remote Sensing and general issues |
| Chair | Sonosuke Fukushima(ENRI) |
| Vice Chair | Toshifumi Moriyama(Nagasaki Univ.) / Akitsugu Nadai(NICT) |
| Secretary | Toshifumi Moriyama(Mitsubishi Electric) / Akitsugu Nadai(ENRI) |
| Assistant | Manabu Akita(Univ. of Electro-Comm.) / Ryo Natsuaki(Univ. of Tokyo) |
| Paper Information | |
| Registration To | Technical Committee on Space, Aeronautical and Navigational Electronics |
|---|---|
| Language | ENG |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Deep Learning for Target Classification from SAR Imagery |
| Sub Title (in English) | Data Augmentation and Translation Invariance |
| Keyword(1) | synthetic aperture radar (SAR) |
| Keyword(2) | automatic target recognition (ATR) |
| Keyword(3) | target classification |
| Keyword(4) | deep learning |
| Keyword(5) | convolutional neural network (CNN) |
| Keyword(6) | data augmentation |
| Keyword(7) | translation invariance |
| Keyword(8) | residual network |
| 1st Author's Name | Hidetoshi Furukawa |
| 1st Author's Affiliation | Toshiba Infrastructure Systems & Solutions Corporation(Toshiba Infrastructure Systems & Solutions) |
| Date | 2017-08-24 |
| Paper # | SANE2017-30 |
| Volume (vol) | vol.117 |
| Number (no) | SANE-182 |
| Page | pp.pp.13-17(SANE), |
| #Pages | 5 |
| Date of Issue | 2017-08-17 (SANE) |