Fig 1.
Encoding and decoding structure of Chinese description of image content.
Fig 2.
Extraction process of global and local image features.
Fig 3.
Global and local image feature fusion.
Fig 4.
Attention weight of image feature.
Fig 5.
Dynamic adjustment mechanism of image feature attention vector.
Fig 6.
Evaluation score of models in validation set.
Table 1.
Evaluation scores of attention mechanism comparison experiment.
Table 2.
Evaluation scores of comparison experiment with other models.
Table 3.
Impact of different beam widths on model evaluation scores.
Table 4.
Comparison of different search methods in this model.
Table 5.
Comparison of different search methods of GLF model.
Table 6.
Subjective comparison of description effects.