In response to the problem of poor detection ability of object detection models for small-scale targets in intelligent transportation scenarios, a fusion method is proposed to enhance the features of small-scale targets, starting from feature utilization and fusion methods. The algorithm is based on the YOLOv4 tiny framework and enhances the utilization of shallow and mid level features on the basis of FPN, improving the detection accuracy of small and medium-sized targets; In view of the problem that the background of the intelligent traffic scene image is cluttered and there is more redundant information, the CBAM attention module is used to improve the attention of the model to the traffic target; To address the problem of data imbalance and prior bounding box adaptation in custom traffic datasets that expand traffic images in COCO and VOC, we propose a Copy Paste method with improved generation method and a K-means algorithm with improved distance measurement to enhance the model's detection ability for corresponding categories. Comparative experiments were conducted on a customized 260 thousand traffic dataset containing public traffic images, and the results showed that compared to YOLOv4 tiny, the proposed algorithm improved mAP by 4.9% while still ensuring the real-time performance of the model.

The most common method for establishing bioequivalence (BE) is to demonstrate similarity of concentration-time profiles in the systemic circulation, as a surrogate to the site of action. However, the similarity, or otherwise, of profiles from two different formulation in the gastrointestinal tract (GIT) may not be associated with same conclusions based on systemic circulation. Here, we have explored the concordance of BE conclusions (or otherwise) for a set of hypothetical formulations based on drug concentration-profiles in various segments of gut vs those in systemic circulation using virtual trials powered by physiologically-based (PBPK) models. Moreover, by changing physiological and biological attributes of the GIT, the impact of Crohn’s Disease on the BE conclusions was explored. The virtual BE (VBE) was applied to an Entocort EC, a pH/time-dependent release formulation of budesonide, as model drug that is targeted at ileum and colon. Substantial “discordance” between the local and systemic outcomes of VBE was observed. Upper GIT segments were much more sensitive to formulation changes than systemic circulation, where the latter led to false conclusions for BE. The ileum and colon showed a lower frequency of discordance with systemic-derived VBE. Nonetheless, a higher f2 value than the default similarity standard (50) was required to achieve local BE. In the case of Crohn’s Disease, a product-specific f2 value might be needed for drugs such as Entocort EC to ensure local BE, knowing the insensitivity of systemic data to show lack of BE. The results, being limited to the single model drug product, Entocort EC, may not provide a a general and conclusive picture. However, for the first time, it demonstrated the potential discordances between the local gut vs systemic BE.