Correction: Li et al. FSDN-DETR: Enhancing Fuzzy Systems Adapter with DeNoising Anchor Boxes for Transfer Learning in Small Object Detection. Mathematics 2025, 13, 287

Error in Table

In the original publication [1], there was a mistake in Table 3. During the final submission, the file for Table 3 was inadvertently overwritten, causing Table 3 to duplicate the contents of Table 2 (COCO results). The corrected

Table 3. Performance comparison of Fuzzy-DETR and baseline models on the AI-TOD-V2 test dataset, evaluated using AP metrics for very tiny ($A{P}_{vt}$), tiny ($A{P}_t$), small ($A{P}_s$), and medium ($A{P}_m$) object sizes.

Basemodel	Method	Backbone	AP	${\mathbf{AP}}_{\mathbf{vt}}$	${\mathbf{AP}}_{\mathit{t}}$	${\mathbf{AP}}_{\mathit{s}}$	${\mathbf{AP}}_{\mathit{m}}$
CNN-based	RetinaNet [25]	ResNet50-FPN	8.92	2.49	8.36	13.06	19.89
	Mask R-CNN [56]	ResNet101-FPN	9.83	1.84	8.76	18.53	23.70
	Faster R-CNN [21]	ResNet101-FPN	13.65	0.04	9.53	25.71	37.84
DETR-like	Deformable DETR [5]	ResNet50	18.86	6.49	17.45	25.03	34.14
	DN-DETR [7]	ResNet50	23.38	9.86	22.60	29.74	38.81
	DINO-DETR [8]	ResNet50	25.88	13.02	25.09	32.02	39.65
	DQ-DETR [12]	ResNet50	29.73	15.09	29.38	36.38	44.64
Ours	Fuzzy-DETR	ResNet50	31.58	18.20	31.51	37.43	46.91

(AI-TOD-V2 results: (APvt), (APt), (APs), and (APm)) appears below. The authors state that the scientific conclusions are unaffected. This correction was approved by the Academic Editor. The original publication has also been updated.

Text Correnction

There was an error in the original publication. One data point was redundantly described. A correction has been made to Section 4, Sub-section 4.6, Paragraph 4:

In addition to small and very small objects, FSDN-DETR demonstrates strong performance across medium objects, with AP_m scores of 46.91, respectively.

The authors state that the scientific conclusions are unaffected. This correction was approved by the Academic Editor. The original publication has also been updated.