YOLO-CIANNA: Galaxy detection with deep learning in radio data - I. A new YOLO-inspired source detection method applied to the SKAO SDC1

Home

All issues

Volume 690 (October 2024)

A&A, 690 (2024) A211

Full HTML

Open Access

Table 5

SDC1 scores and related properties for source catalogs from different teams and methods.

Team (method)	M_s (Score)	N_det	N_match	N_false	N_bad ∈ N_false	Purity	$\bar{s}$ ${\bar s}$
Post-challenge results

MINERVA (YOLO-CIANNA)	480450	724480	680000	44480	16839	93.86%	0.7719
↪ purity-focus thresholds	418434	541 542	536412	5130	2506	99.06%	0.7896
JLRAT2 (JSFM2)	298201	502 146	484212	17 934	2274	96.43%	0.6529

Original challenge results

Engage-SKA (PROFOUND)	200939	421992	418384	3608	2677	99.15%	0.4889
Shanghai (multiple methods)	158841	292646	291 553	1093	698	99.63%	0.5486
ICRAR (CLARAN)	142784	279 898	259806	20092	6875	92.82%	0.6269
…	…	…	…	…	…	…	…

Notes. The bold element represents the target metric that is optimized for each result.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.