Method for detecting multiple targets, electronic equipment and storage medium

The embodiment of the invention relates to the technical field of target detection, and particularly relates to a method for training a multi-target detection model, electronic equipment and a storagemedium. The type of a target corresponding to each image is determined according to a real label of each image, so that a loss function corresponding to each image is determined, the error interference among different types of targets can be effectively eliminated, and a more accurate multi-target detection model is obtained. Furthermore, in order to realize subdivision category positioning detection of different types of targets, image annotation does not need to be carried out again for each pattern sample subset, i.e., subdivision categories and positions of various types of targets do notneed to be annotated on the same image, so that the labor cost can be saved. 本发明实施例涉及目标检测技术领域，尤其涉及一种训练多目标检测模型的方法、电子设备及存储介质，根据各图像的真实标签，确定各图像对应的目标的类型，从而，确定各图像对应的损失函数，可以有效消除不同类型的目标之间的误差干扰，得到更加准确的多目标检测模型。此外，为了实现各不同类型的目标的细分类别定位检测，