Chair Manufacturing Frame Dataset

This dataset consists of images of various chair frames, primarily metallic structures, captured from different angles and positions. The dataset is being developed for public access to facilitate research and development in pose estimation technologies in our further studies. The chairs present a unique challenge for pose detection algorithms due to their structural uniformity and repetitive patterns. To enhance the dataset’s utility, each image is captured under controlled lighting conditions, with attention to minimizing noise and occlusions that could affect the accuracy of pose estimation models. The dataset includes multiple instances of the same chair model with varying degrees of rotation and translation, providing a comprehensive basis for training convolutional neural networks (CNNs) and other machine learning algorithms commonly used in pose estimation tasks. The inherent symmetries in the chair frames create a demanding testbed for feature extraction and keypoint detection, as traditional algorithms may struggle to distinguish between similar geometric structures. In our studies, we aim to incorporate advanced techniques such as 3D point cloud generation and use of depth maps to improve the precision of pose detection, leveraging both monocular and stereo vision systems. Additionally, we plan to explore the use of synthetic data augmentation methods to increase the dataset's diversity and robustness, which is critical for developing generalizable models in real-world applications. This dataset will not only support pose estimation research but will also contribute to broader studies in object recognition, segmentation, and tracking in dynamic environments. The upcoming scientific research in out lab will focus on "Chair Pose Estimation," aiming to leverage machine learning and computer vision techniques to accurately detect and model the pose of chairs in different environments. This research has practical applications in automation, robotic interaction with furniture, and augmented reality. The dataset will be made publicly available, enabling the academic and industrial community to conduct further research and develop solutions in pose estimation and related fields.