Improving work detection by segmentation heuristics pre-training on factory operations video.

The measurement of work time for individual tasks by using video has made a significant contribution to a framework for productivity improvement such as value stream mapping (VSM). In the past, the work time has been often measured manually, but this process is quite costly and labor-intensive. For these reasons, automation of work analysis at the worksite is needed. There are two main methods for computing spatio-temporal information: by 3D-CNN, and by temporal computation using LSTM after feature extraction in the spatial domain by 2D-CNN. These methods has high computational cost but high model representational power, and the latter has low computational cost but relatively low model representational power. In the manufacturing industry, the use of local computers to make inferences is often required for practicality and confidentiality reasons, necessitating a low computational cost, and so the latter, a lightweight model, needs to have improved performance. Therefore, in this paper, we propose a method that pre-trains the image encoder module of a work detection model using an image segmentation model. This is based on the CNN-LSTM structure, which separates spatial and temporal computation and enables us to include heuristics such as workers' body parts and work tools in the CNN module. Experimental results demonstrate that our pre-training method reduces over-fitting and provides a greater improvement in detection performance than pre-training on ImageNet.

Radiographic deformities of the lower extremity in patients with spontaneous osteonecrosis of the knee.

BACKGROUND: Spontaneous osteonecrosis of the knee (SONK) is one of the acute knee pain disorders arising in elderly patients. The presence of knee varus alignment and the size of necrotic area have been reported as the negative prognostic factors in prior studies. However, no previous study has yet clarified the radiological analysis of the lower extremity in SONK compared with that in osteoarthritis. The purpose of this study was therefore to identify the radiographic findings of the lower extremity in SONK. METHODS: Sixty-three knees of Kellgren-Lawrence classification grade 1 or 2 without any trauma treated between April 2012 and March 2014 were enrolled in this study. These knees were divided into two groups according to their magnetic resonance imaging (MRI) findings: SONK group (31 knees) and OA group (32 knees). Using a long leg standing X-ray, femorotibial angle (FTA), mechanical axis deviation (MAD), mechanical lateral distal femoral angle (mLDFA), medial proximal tibial angle (MPTA) and joint line convergent angle (JLCA) were compared between groups. Correlation between each parameter and the width ratio (WR) of the necrotic lesion were analyzed. RESULTS: FTA, MAD, MPTA and JLCA showed significant differences between the SONK and OA groups. In the SONK group, FTA was positively correlated with WR, and, MAD and MPTA was negatively correlated with WR. CONCLUSIONS: Compared with OA, SONK is associated with a significantly larger varus deformity at the proximal tibia, and larger joint play in the coronal plane.