Unverified50% confidenceFactExact time
目标检测传统上由 YOLO、Faster R-CNN、DETR 等专用架构主导，通过专用检测头回归边界框坐标与类别概率
1
Sources
50%
Confidence
Long-term
Relevance
7/11/2026
First Seen
Sources
微调PaliGemma 2实现自定义目标检测：从通用到专用的实践指南
redditr/learnmachinelearning7/10/2026
Related Claims
UnverifiedOne-stage检测器（如YOLO、SSD）将候选框生成与分类合并为一步，在单次前向传播中完成所有预测73% similarVerifiedThe YOLO series of object detection models has evolved from V1 to V11.69% similarUnverified自动标注管线常见架构是将目标检测模型（如 YOLO 系列、DINO、Grounding DINO）与 SAM 级联，先产出边界框再作为提示输入 SAM 获取精细掩膜68% similarUnverifiedFor resource-constrained graduation project scenarios, it is recommended to treat aerial human body detection as the core deliverable of the project, fine-tuning YOLO or DETR on the SARD/HERIDAL datasets67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/486018API
curl https://kongchang.com/api/v1/knowledge/claims/486018MCP
get_claim(id=486018)