Fig. 11From: 3D human pose estimation model using location-maps for distorted and disconnected images by a wearable omnidirectional cameraThe L2 loss transition on the validation dataset-A during fine-tuning for the different scale models(1x, 2x, and 4x). The 4x-scale and 2x-scale model fall into a local solution after approximately 9 and 10 epochsBack to article page