| Presentation | 2022-06-10 [Tutorial Lecture] How to build a High-Precision and Efficient Robot Vision: Dataset Generation and Hardware Implementation for Deep Learning Hakaru Tamukoh, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | This tutorial lecture explains a construction method for high-precision and efficient robot vision that includes a semi-automatic dataset generation method and an implementation method of deep neural networks (DNNs) on field-programmable gate arrays (FPGAs). The proposed dataset generation method ultimately reduces the time-consuming manual annotation process, and a generated dataset for DNNs can be prepared in about two hours. I show the evaluation results for a DNN trained by the generated dataset under real-world conditions through robot competitions such as RoboCup and World Robot Summit. We also show an example of FPGA implementation of YOLO v3 tiny. Through this presentation, I show examples of solutions for dataset preparation and power-consumption issues, which are significant barriers to edge applications of artificial intelligence. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Deep Learning / DNN / Dataset / FPGA / Robot Vision / RoboCup / World Robot Summit |
| Paper # | SIS2022-10 |
| Date of Issue | 2022-06-02 (SIS) |
| Conference Information | |
| Committee | SIS / IPSJ-AVM |
|---|---|
| Conference Date | 2022/6/9(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | KIT(Wakamatsu Campus) |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Intelligent Multimedia Systems, Applied Embedded Systems, Three-Dimensional Image Technology (3DIT), etc. |
| Chair | Noriaki Suetake(Yamaguchi Univ.) / Sei Naito(KDDI Research, Inc.) |
| Vice Chair | Tomoaki Kimura(Kanagawa Inst. of Tech.) / Naoto Sasaoka(Tottori Univ.) |
| Secretary | Tomoaki Kimura(NTT) / Naoto Sasaoka(National Inst. of Tech., Ube College) / (NTT) |
| Assistant | Soh Yoshida(Kansai Univ.) / Yoshiaki Makabe(Kanagawa Inst. of Tech.) |
| Paper Information | |
| Registration To | Technical Committee on Smart Info-Media Systems / Special Interest Group on Audio Visual and Multimedia Information Processing |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | [Tutorial Lecture] How to build a High-Precision and Efficient Robot Vision: Dataset Generation and Hardware Implementation for Deep Learning |
| Sub Title (in English) | |
| Keyword(1) | Deep Learning |
| Keyword(2) | DNN |
| Keyword(3) | Dataset |
| Keyword(4) | FPGA |
| Keyword(5) | Robot Vision |
| Keyword(6) | RoboCup |
| Keyword(7) | World Robot Summit |
| 1st Author's Name | Hakaru Tamukoh |
| 1st Author's Affiliation | Kyushu Institute of Technology(Kyutech) |
| Date | 2022-06-10 |
| Paper # | SIS2022-10 |
| Volume (vol) | vol.122 |
| Number (no) | SIS-62 |
| Page | pp.pp.45-48(SIS), |
| #Pages | 4 |
| Date of Issue | 2022-06-02 (SIS) |