唐澤 拓己
Takumi Karasawa
CV Researcher
/ AI Group Manager
EVERSTEEL Inc.
Interest
Machine Learning, Computer Vision, Object Detection, Domain Adaptation, Few-shot Learning
Experience
Apr. 2022 - now: AI Group Manager at EVERSTEEL Inc.
Apr. 2019 - July 2023: AI (Computer Vision) Research Engineer at DeNA Co., Ltd. / GO Inc.
Feb. 2021 - March. 2021: [freelance] Computer Vision Researcher at Konica Minolta, Inc.
Feb. 2020 - March. 2020: [freelance] Computer Vision Researcher at NTT DATA
Education
Bachelor's and Master's Graduate from the Harada Lab. (MIL),
Mechano-Informatics, Graduate School of Information Science and Technology, the University of Tokyo.Ph.D. Student at the Kawakami Lab. (Open-vision lab.),
School of Engineering, Tokyo Institute of Technology.
Papers
Takumi Karasawa, Nakamasa Inoue, Rei Kawakami, "Spatiality-Aware Prompt Tuning for Few-shot Small Object Detection.", The 2024 IEEE International Conference on Image Processing (ICIP), 2024. 【Best Student Paper Award (2nd runner-up)】
田中大智, 唐澤拓己, 川上玲, “鉄スクラップのリサイクルのための生成モデルを活用したテキストと画像の少数例示による異常の領域分割”, The 27th Meeting on Image Recognition and Understanding (MIRU), 2024.
Hiroto Honda, Motoki Kimura, Takumi Karasawa, Yusuke Uchida. "End-to-End Monocular Vanishing Point Detection Exploiting Lane Annotations." arXiv preprint arXiv:2108.13699, 2021.
Takumi Karasawa*, Kohei Watanabe*, Qishen Ha, Antonio Tejero-De-Pablos, Yoshitaka Ushiku, Tatsuya Harada. "Multispectral Object Detection for Autonomous Vehicles." Thematic Workshop of 25th Annual ACM International Conference on Multimedia (ACMMM), 2017.
Qishen Ha, Kohei Watanabe, Takumi Karasawa, Yoshitaka Ushiku, Tatsuya Harada. "MFNet: Towards Real-Time Semantic Segmentation for Autonomous Vehicles with Multi-Spectral Scenes." The 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2017.
Kaggle competitions
all results:
Nov. 2021: NFL Health & Safety - Helmet Assignment:16/825, Top 2% (Silver medal)
Nov. 2020: Lyft Motion Prediction for Autonomous Vehicles:28/935, Top 3% (Silver medal)
Sep. 2019: APTOS 2019 Blindness Detection:79/2943, Top 3% (Silver medal)
June 2019: Freesound Audio Tagging 2019:23/880, Top 3% (Silver medal)
National Qualifications
Dec. 2015: Applied Information Technology Engineer
June 2016: Information Security Specialist
Talks / Events
July 2024: ㈱ 鉄リサイクリング・リサーチ主催 第4回鉄スクラップセミナー招待講演「AIで作る鉄鋼業界のサーキュラーエコノミー 」
Feb. 2023: DeNA、MoT 合同 AI 技術共有会「Face Recognition @ ECCV2022」
May 2022: DeNA、MoT 合同 AI 技術共有会「顔認識(face recognition)と最近のArcFaceまわりと」
Oct. 2021: Discovery DataScience Meet up (DsDS) #2 招待講演「交通環境におけるコンピュータビジョン技術実応用 」
July 2021: DeNA、MoT 合同 AI 技術共有「Monocular 3D Object Detection @CVPR2021」
Apr. 2021: MoT CV輪講「今日から始める PyTorch Lightning」
Mar. 2021: 第42回 IBISML研究会 「機械学習の実世界への適用」セッション招待講演「交通環境におけるコンピュータビジョン技術実応用:DRIVE CHART 」
July 2020: 第3回 全日本コンピュータビジョン勉強会, CVPR2020読み会「Self-supervised Monocular Scene Flow Estimation」(youtube)
July 2020: DeNA、MoT 合同 AI 技術共有会「CVPR 2020 参加レポート」
July 2020: DeNA、MoT 合同 AI 技術共有会「Domain Adaptation 入門」
May 2020: MoT CV輪講「Semi-supervised learning かじってみた」
Mar. 2020: DeNA TechCon 2020「コンピュータビジョン技術の実応用 〜DRIVE CHARTにおける脇見・車間距離不足検知〜」(youtube)
Dec. 2019: DeNA AI 技術共有会「ICCV 2019 論文紹介」
Dec. 2019: DeNA20周年イベント「1日DeNA学校:AI入門」(日経新聞)
Dec. 2019: DeNA AI 技術共有会「Rethinking ImageNet Pre-training」
Oct. 2019: DeNA CV輪講「SelFlow: Self-supervised Learning of Optical Flow」
Articles
Aug. 2019: DeNA Engineer's Blog「コンピュータビジョン最新論文調査 Segmentation 編」
Aug. 2019: DeNA Engineer's Blog「コンピュータビジョン最新論文調査 Object Tracking 編」
Contact
zaki19930927[at]gmail.com
takumi.karasawa[at]eversteel.co.jp