Amharic Language Image Captions Generation Using Hybridized Attention-Based Deep Neural Networks

Table 3

The models performance on Flickr8k and BNATURE datasets.

Dataset	Model	1G-BLEU	2G-BLEU	3G-BLEU	4G-BLEU
BNATURE	CNN-Bi-GRU	42.6	27.9	23.6	16.4
	Hybridized model	63.1	53.3	47.4	42.7
Flickr8k English	Bag-LSTM	59.7	41.4	28.1	18.2
	Hybridized model	65.6	53.2	45.3	39.4