In response to the increasing prevalence of dental diseases, dental health, a vital aspect of human well-being, warrants greater attention. Panoramic X-ray images (PXI) and Cone Beam Computed Tomography (CBCT) are key tools for dentists in diagnosing and treating dental conditions. Additionally, deep learning for tooth segmentation can focus on relevant treatment information and localize lesions. However, the scarcity of publicly available PXI and CBCT datasets hampers their use in tooth segmentation tasks. Therefore, this paper presents a multimodal dataset for Semi-supervised Tooth Segmentation (STS-Tooth) in dental PXI and CBCT, named STS-2D-Tooth and STS-3D-Tooth. STS-2D-Tooth includes 4,000 images and 900 masks, categorized by age into children and adults. Moreover, we have collected CBCTs providing more detailed and three-dimensional information, resulting in the STS-3D-Tooth dataset comprising 148,400 unlabeled scans and 8,800 masks. To our knowledge, this is the first multimodal dataset combining dental PXI and CBCT, and it is the largest tooth segmentation dataset, a significant step forward for the advancement of tooth segmentation.

Segment anything model (SAM), a vision foundation network trained on a massive segmentation corpus, exhibits a superior boundary localization capability for nature images. This work aims to leverage such strengths to develop a deep unsupervised edge detection (UED) framework for alleviating the high reliance on dense labeling. However, applying vanilla SAM to edge detection fails to identify the salient edge cues but only the semantic boundary. This article introduces a lightweight adapter-tuning scheme to learn detailed edge information for filling the gap between boundary and edge, enabling a well-fitting even with limited training data. Moreover, considering the low-quality pseudo labels used in our UED framework, we propose two training strategies, adaptive progressive learning and gradient-guided pseudo label updating, to alleviate the impact of noisy labels from traditional UED methods. Extensive experiments demonstrate that our method achieves comparable results to state-of-the-art fully supervised edge detectors.

Compared to models pre-trained on ImageNet, the segment anything model (SAM) has been trained on a massive segmentation corpus, excelling in both generalization ability and boundary localization. However, these strengths are still insufficient to enhance shadow detection without additional training, and it raises the question: do we still need precise manual annotations to fine-tune SAM for high detection accuracy? This paper proposes an annotation-free framework for deep unsupervised shadow detection (USD) by leveraging SAM’s capabilities. The key lies in how to exploit the abilities acquired from a large-scale corpus and utilize them to improve downstream tasks. Instead of directly fine-tuning SAM, we propose a prompt-like tuning method to inject task-specific cues into SAM in a light-weight manner, namely ShadowSAM. This adaptation manner can ensure a good fitting when training data is limited. Moreover, considering that the pseudo labels used in our framework are generated by traditional USD approaches and may contain severe label noises, we propose an illumination and texture-guided updating strategy to selectively boost the quality of pseudo masks. To further improve the model’s robustness, we design a mask diversity index to establish easy-to-hard subsets for incremental curriculum learning. Extensive experiments on benchmark datasets ( i.e ., SBU, UCF, ISTD, and CUHK-Shadow) demonstrate that our unsupervised solution can achieve comparable performance to state-of-the-art (SOTA) fully supervised methods. Our code is available at this repository.