Object detection is one of the core tasks of computer vision, and bounding box (bbox) regression is one of the basic tasks of object detection. In recent years of related research, bbox regression is often used in the Intersection over Union (IoU) loss and its improved version. In this paper, for the first time, we introduce the Dice coefficient into the regression loss calculation and propose a new measure which is superior to and can replace the IoU. We define three properties of the new measure and prove the theory by mathematical reasoning and analysis of the existing work. This paper also proposes the N-IoU regression loss family. And the superiority of the N-IoU regression loss family is proved by designing simulation experiments and comparative experiments. The main results of this paper are: (1) The proposed new measure is better than IoU which can be used to evaluate bounding box regression, and the three properties of the new measure can be used as a broad criterion for the design of regression loss functions; and (2) we propose N-IoU loss. The parameter n of N-IOU can be debugged, which can be widely adapted to different application scenarios with higher flexibility, and the regression performance is better.

Vision transformers (ViTs) have become one of the dominant frameworks for vision tasks in recent years because of their ability to efficiently capture long-range dependencies in image recognition tasks using self-attention. In fact, both CNNs and ViTs have advantages and disadvantages in vision tasks, and some studies suggest that the use of both may be an effective way to balance performance and computational cost. In this paper, we propose a new hybrid network based on CNN and transformer, using CNN to extract local features and transformer to capture long-distance dependencies. We also proposed a new feature map resolution reduction based on Discrete Cosine Transform and self-attention, named DCT-Attention Down-sample (DAD). Our DctViT-L achieves 84.8% top-1 accuracy on ImageNet 1K, far outperforming CMT, Next-ViT, SpectFormer and other state-of-the-art models, with lower computational costs. Using DctViT-B as the backbone, RetinaNet can achieve 46.8% mAP on COCO val2017, which improves mAP by 2.5% and 1.1% with less calculation cost compared with CMT-S and SpectFormer as the backbone.

Fine-grained unmanned aerial vehicle (UAV) type classification benefits from complementary sensors, yet real deployments face incomplete observations and heterogeneous stream reliability. We present a missing-aware fusion framework for RGB detections (YOLO11-S), Livox Avia LiDAR, 360° LiDAR, and radar-enhanced point clouds on MMAUD. Confidence–density scores are masked and renormalized so fusion weight is assigned only to observed streams (Eq. (2)); each absent stream still occupies a classifier slot via a learnable missing embedding that is not multiplied by the zero weight (Eq. (1)), keeping absence as a trainable signal rather than a numerical zero. On a time-stratified split, Tiny MLP (≈ 0.1 M fusion parameters) reaches 96.12% test accuracy in a representative seed-42 run. Over seeds 42–46, final test accuracy is 95.29 ± 0.60% without training-time modality dropout and 95.77 ± 0.90% with 30% training ModDrop (mean ± sample s.d.). Under matched 30% test missingness, accuracy remains 83.68% (seed 42). Protocol-aligned baselines and orthogonal ablations support the slot rule. Fusion-stage inference is about 2.1 ms per sample with pre-extracted features on an RTX 4090. Under our fixed split and frozen features—not as a challenge-leaderboard claim—compact missing-aware fusion is practical for heterogeneous UAV sensing under partial sensor availability.