Loss functions is a crucial factor that affects the detection precision in object detection task. In this paper, we optimize both the loss functions for classification and localization simultaneously. Firstly, by multiplying an IoU-based coefficient by the standard cross entropy loss in standard classification loss function, the correlation between localization and classification is established. Compared to the existing studies, in which the correlation is only applied to improve the localization accuracy for positive samples, this paper utilizes the correlation to obtain the really hard negative samples and decrease the misclassified rate for negative samples remarkably. Besides, a novel localization loss named MIoU is proposed by incorporating a Mahalanobis distance between predicted box and target box, which eliminate the gradients inconsistency problem in DIoU loss, further improving the localization accuracy. Finally, sufficient experiments on nighttime vehicle detection have been done. Our results show than when train with the proposed loss functions, the detection performance can be outstandingly improved.