Tablet Recognition and Labelling

Tablet recognition relies heavily on physical attributes such as color, size, and shape. However, these features can be influenced by environmental conditions, leading to variations that may cause recognition errors. Such discrepancies can result in medication errors due to damaged labels, incorrect identification, or mismatched intake, potentially putting patients at risk. This report introduces a trained recognition system developed using Keras and Tensor Flow to facilitate the rapid and accurate labeling of various tablets. The system identifies tablets through object recognition and links them to a database to retrieve their names and relevant details. Upon detection, a pre-trained dataset is utilized to confirm the tablet’s identity. Additionally, the dataset contains use cases and essential information for each tablet. The proposed solution supports automated medicine identification, and its performance has been validated through experimental results.