@article{oai:ipsj.ixsq.nii.ac.jp:00224352, author = {森茂, 智彦 and 山口, 穂高 and 藤巻, 吾朗 and 生駒, 晃大 and Tomohiko, Morimo and Hodaka, Yamaguchi and Goroh, Fujimaki and Akihiro, Ikoma}, issue = {2}, journal = {情報処理学会論文誌}, month = {Feb}, note = {棚卸作業において,画像認識によって部材の数量を求めることは有用である.しかし,その精度が100%となることは保証されないため,正確な数量の把握には目視による認識結果の修正も必要である.当然この修正は速く正確に行えることが望まれる.そこで本論文では,画像を格子線で区切ることによって修正作業が効率的に行えるとの仮説を立て,検証する.木製の丸棒部材を用いて検証した結果,未検出の部材のみが含まれる画像の修正では格子線の影響が限定的であり,誤検出の印を消す修正および未検出と誤検出の同時修正が必要な画像において格子線を用いた条件で未修正数が減少した.一方で,修正に要した作業時間は格子線の有無によって変わるとはいえなかった.よって,丸棒部材において格子線は,画像認識後の修正作業を阻害する可能性は少なく,誤検出した印の修正に有効であると結論づけられる., In inventory work, determining the number of parts and materials by image recognition is useful. However, since 100% accuracy is not guaranteed, visual correction of the recognition results is also necessary to accurately determine the quantity. This correction should be fast and accurate. This study hypothesizes and tests that the correction work can be done efficiently by dividing the image by grid lines. The results of the verification using wooden round bars members showed that the effect of grid lines was limited in the correction of images containing only undetected members and that the number of uncorrected images decreased under the condition of using grid lines in images requiring correction to erase false-positive marks and simultaneous correction of undetected and false-positive marks. Alternatively, the working time required for the correction did not vary depending on the presence or absence of grid lines. Therefore, the grid lines are less likely to interfere with the correction process after image recognition in round bars members and are effective in correcting false-positive marks.}, pages = {289--300}, title = {円形集合体検出後の修正作業における格子線の効果:木製丸棒部材の在庫画像を用いた検証}, volume = {64}, year = {2023} }