. 2018 Jun 14;8:9145. doi: 10.1038/s41598-018-27272-w

Table 1.

Performance metrics of algorithm on spring 2016 test sets, which are the ‘less-cluttered’ 24 sets labeled {A1, …, A24} having 50 frames, ‘high-cluttered’ group containing 24 sets labeled {B1, …, B24} of 50 frames and the aggregate performance of all the total 2400 testing images.

Adapted metric	Formulae	‘Less-cluttered’ group	‘High-cluttered’ group	Aggregate
Average detection accuracy	$\frac{\sum a l g o r i t h m c o u n t}{\sum h u m a n c o u n t + e r r o r m a r g i n}$	$\frac{2036}{2150} = 94.70 %$	$\frac{459}{475} = 97.00 %$	$\frac{2495}{2625} = 95.05 %$
Average alarm-to-egg ratio	$\frac{\sum e x c e s s c o u n t (a f f e c t e d s e t s)}{\sum h u m a n c o u n t + e r r o r m a r g i n}$	$\frac{16}{2150} = 0.75 %$	$\frac{21}{475} = 4.00 %$	$\frac{37}{2625} = 1.40 %$
Average miss-to-egg ratio	$\frac{\sum u n d e r c o u n t s (a f f e c t e d s e t s)}{\sum h u m a n c o u n t + e r r o r m a r g i n}$	$\frac{13}{2150} = 0.60 %$	$\frac{3}{475} = 0.63 %$	$\frac{16}{2625} = 0.61 %$
Average precision	$\frac{\sum a l g o r i t h m c o u n t}{\sum h u m a n c o u n t + e r r o r m a r g i n + \sum e x c e s s c o u n t}$	$\frac{2036}{2166} = 94.00$	$\frac{459}{496} = 93.00 %$	$\frac{2495}{2662} = 93.73 %$
F₁- Score	$\frac{2 \times A v e r a g e p r e c i s i o n \times A v e r a g e d e t e c t i o n a c c u r a c y}{A v e r a g e p r e c i s i o n + A v e r a g e d e t e c t i o n A c c u r a c y}$	0.943	0.949	0.944

Error margin is found by taking 5% (i.e., the upper bound of the error bar) of the total human count for all image sets in each category.