| Presentation | 2020-03-17 Semi-Supervised Temporal Segmentation of Industrial Operation Video based on Deep Metric Learning Daiki Kawamori, Kazuaki Nakamura, Naoko Nitta, Noboru Babaguchi, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Today, cameras are often installed in many production sites for various purposes. However, untrimmed raw videos captured by the cameras are hard to use. Hence, it is desired to automatically segment the videos along the time axis and recognize which kind of operation is performed in each segment. We call this task ``temporal segmentation,'' which is useful for making an operational record and building a new plan. To achieve high performance of temporal segmentation, we have to use effective video features. Such features can hardly be obtained by unsupervised learning, whereas supervised learning has a drawback that collecting a lot of training data is labor-intensive. From these backgrounds, in this paper, we propose a method of obtaining effective features based on semi-supervised distance metric learning, under the assumption that only a few frames in input industrial operation videos are labeled and given as training data. To achieve high performance, the proposed method automatically build a tree structure that represents hierarchal relationship between class labels, and separately obtain an effective feature for each branch in the tree. In our experimental results, we achieved the temporal segmentation performance of 0.956 on the F measure, even when less than 3% of all frames in the input videos are labeled. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | industrial operation video / deep metric learning / temporal segmentation / semi-supervised clustering / label hierarchy tree |
| Paper # | PRMU2019-92 |
| Date of Issue | 2020-03-09 (PRMU) |
| Conference Information | |
| Committee | PRMU / IPSJ-CVIM |
|---|---|
| Conference Date | 2020/3/16(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | Yoichi Sato(Univ. of Tokyo) |
| Vice Chair | Toru Tamaki(Hiroshima Univ.) / Akisato Kimura(NTT) |
| Secretary | Toru Tamaki(NTT) / Akisato Kimura(OMRON SINICX) |
| Assistant | Yusuke Uchida(DeNA) / Takayoshi Yamashita(Chubu Univ.) |
| Paper Information | |
| Registration To | Technical Committee on Pattern Recognition and Media Understanding / Special Interest Group on Computer Vision and Image Media |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Semi-Supervised Temporal Segmentation of Industrial Operation Video based on Deep Metric Learning |
| Sub Title (in English) | |
| Keyword(1) | industrial operation video |
| Keyword(2) | deep metric learning |
| Keyword(3) | temporal segmentation |
| Keyword(4) | semi-supervised clustering |
| Keyword(5) | label hierarchy tree |
| 1st Author's Name | Daiki Kawamori |
| 1st Author's Affiliation | Osaka University(Osaka Univ.) |
| 2nd Author's Name | Kazuaki Nakamura |
| 2nd Author's Affiliation | Osaka University(Osaka Univ.) |
| 3rd Author's Name | Naoko Nitta |
| 3rd Author's Affiliation | Osaka University(Osaka Univ.) |
| 4th Author's Name | Noboru Babaguchi |
| 4th Author's Affiliation | Osaka University(Osaka Univ.) |
| Date | 2020-03-17 |
| Paper # | PRMU2019-92 |
| Volume (vol) | vol.119 |
| Number (no) | PRMU-481 |
| Page | pp.pp.139-144(PRMU), |
| #Pages | 6 |
| Date of Issue | 2020-03-09 (PRMU) |