Artificial Intelligence (AI) is the science of making computers perform tasks that conventionally require human intelligence [1]. AI has recently seen significant growth due to the availability of large digital datasets and rapid increase in computational power [2].

Machine Learning (ML) and Deep Learning (DL) are integral subsets of AI. ML involves creating algorithms that can automatically learn patterns and make decisions from data, often requiring human-designed feature extraction. On the other hand, DL is a specialized form of ML that enables machines to autonomously discover essential patterns and features for tasks like detection and classification [3]. DL employs multiple layers of non-linear modules to transform raw data into increasingly abstract representations, proving particularly effective in tasks such as image classification and segmentation [4].

The use of AI in dentistry is experiencing a surge in popularity, with the development of several platforms intended to facilitate the diagnosis of dental conditions and aid in treatment planning. Those platforms depend on the concepts of ML and DL which have numerous applications in oral radiology, including caries detection [5], bone loss detection [6], apical lesion detection [7], and root fracture [8]. Therefore, these tools are rapidly becoming indispensable instruments in the field of oral health imaging, offering numerous opportunities for enhancing diagnostic precision, streamlining workflow, and improving patient care [9], [10], [11], [12].

Deep learning in imaging involves three key tasks: segmentation, object detection and classification. Object detection defines rectangular Regions of Interest (ROIs) to locate objects of interest, while segmentation essentially draws a clear distinction between the objects of interest and the background through labeling precise pixels, hence separating objects from backgrounds. Object detection provides a rough location, while segmentation offers precise object delineation. These tasks are chosen based on the specific imaging application's requirements and complement each other in addressing different objectives. Deep learning in imaging also encompasses a third crucial task: classification. DL systems can aid in both detection and classification, serving a twofold purpose. Firstly, they determine whether an image contains a lesion or anomaly, effectively flagging images that need expert evaluation. Secondly, these systems can classify or categorize the identified anomalies, providing insights into the most probable diagnosis [13].

Despite the increasing interest in this field, several factors hinder the effective deployment of DL models [14,15]. These challenges encompass the necessity of advanced DL knowledge and programming skills, dataset preparation requirements, significant computational demands, and intricate algorithms [16,17]. Furthermore, many of the concepts proposing the integration of DL in dentistry originate from dental researchers, clinicians, or dentists, who often possess limited coding backgrounds. Consequently, the majority of end-users lack the opportunity to independently develop their ideas without substantial external assistance [18]. These complexities often result in the centralization of research and AI applications within a limited number of entities. This prevents a broader dissemination of AI-related applications and research opportunities [19].

No-code computer vision is among the latest breakthroughs in AI that have been developed to overcome the limited coding knowledge. This advancement empowers end-users to develop, train, and test their own AI models, without requiring them to possess extensive expertise in coding or software engineering [20]. No-code computer vision models are developed by annotating the spatial features of an object on multiple images to create an algorithm, without coding. Spatial features may consist of the relationships and arrangements of pixels in an image. Spatial features help users identify the object accurately, as well as its location and relation with other objects in an image. Subsequently, this automated algorithm can detect these features within new images, potentially streamlining the practical aspects of deploying, managing, monitoring, and sustaining ML models in real-world scenarios [21]. By adopting this approach, more professionals can gain the ability to harness the potential of machine learning for diverse applications [22,23].

No-code computer vision has been acknowledged in the literature by other names, such as; code-free machine learning [24], code-free deep learning [25], automated deep learning [26], and automated machine learning [27].The low-code computer vision term have also been reported in the literature. However, unlike no-code platforms, low-code platforms involve some degree of coding but aim to minimize the amount of manual coding required [28].No-code computer vision has been applied and tested in the medical field in diagnosing a diverse range of diseases from medical imaging, including the detection of diabetic retinopathy and open-angle glaucoma using retinal images [18]. It resulted in comparable results to conventional trusted deep learning algorithms and ground truth. However, the authors are not aware of studies that have evaluated the utility of no-code computer vision models in the field of dentistry.

This study aimed to develop, train, and evaluate a no-code model specifically designed for the segmentation of dental restorations on panoramic radiographs using a no-code computer vision platform. The null hypothesis was that the no-code AI model will not be able to differentiate between restorations and other structures on the panoramic radiographs.