週報(HU Mengya)

Weekly Report【2025/12/22– 12/28】
1. Master’s Thesis:
Designed a dataset acquisition plan, manually selected broken rice and chalky rice samples, and captured a randomly mixed dataset of whole and broken rice grains. Organized the broken rice dataset and removed low-quality images. Conducted trial shoots of chalky rice and cracked rice grains to explore the optimal imaging ways.

2. Japanese:
Memorizing one-minute NHK news, attended Japanese classes, completed homework, and memorizing vocabulary. Printed the kanji textbook and began memorizing kanji step by step.

週報(中村)12/22

現在の課題

・蛇口と手の距離が近いとき、洗うつもりがなくてもON判定が出る(蛇口の上や横で作業しているとき)

解決策

・蛇口を認識させる

→蛇口が認識できない時は、水を出さないようにする。       蛇口を覚えさせる。

・カメラの位置を真上ではなく、斜めからの画角にずらす。

→現在は蛇口の先端に座標点をうち、その点と手の距離で判定を行っている。

週報(HU Mengya)

Weekly Report【2025/12/08– 12/21】

1. Journal Paper: Rewrote Chapters 1 and 2, added an additional comparative experiment, and revised the analysis of the ablation experiments.
2. Master’s Thesis: Reconsidered the thesis topic and formulated the outline, technical route, and innovations.
3. Japanese: Attended classes, completed homework, memorized vocabulary, and practiced memorizing news scripts.

週報(SUN YUYA)

  1. This week, my main research progress is as follows:

    1. I am revising the manuscript submitted to Pattern Recognition, focusing on improving the paper structure and addressing the reviewers’ comments. In parallel, I am updating and rerunning the corresponding experiments to strengthen the experimental validation.

    2. I am also working on the preparation of the manuscript for ICIAE. This includes refining the methodology description and implementing the related experiments to support the proposed approach.

    I will continue to refine both manuscripts and report further progress in the coming week.

週報(飯田)

キャリブレーションした
・yoloで検出した建物ボックスの中から縦線を選択(手動)
・同様に検出した電柱のトップとボトムを検出し方向線を抽出(自動)
・角度を計算(自動)

理論値を求めるのは難しいので、
類似研究の「電柱のキーポイントによる傾き推定」を再現して、比較することで、有用性を示していきたい。(特にパースがかかった画面端の電柱)