Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386324079> ?p ?o ?g. }
Showing items 1 to 58 of
58
with 100 items per page.
- W4386324079 abstract "Many traditional image recognition technologies are based on large-scale calculation, and there is an irreconcilable contradiction between the amount of calculation and the accuracy of calculation. In recent years, the development of ANN (artificial neural network) technology provides a new way to solve this problem. This paper focuses on the design of image recognition and generation algorithm based on ANN. This paper proposes an image recognition model based on BPNN (BP neural network), and introduces a multi attention mechanism into the language decoding model. By extracting image features at multiple levels and connecting image features at different levels to the multi-level language model through multi attention structure, the language decoding network can adaptively determine the weight of image features at each level used in generating description. The research results show that the improved BP algorithm is slightly better than the standard BP algorithm. It shows that neural network ensemble overcomes the instability of a single neural network and has better generalization ability. The multi-scale feature fusion model with increased attention mechanism generates more accurate image description statements than the model without increased attention mechanism, and the evaluation index is improved to some extent. At the same time, information such as the number and color of objects in the image can be obtained from details." @default.
- W4386324079 created "2023-09-01" @default.
- W4386324079 creator A5021355100 @default.
- W4386324079 date "2023-06-01" @default.
- W4386324079 modified "2023-09-27" @default.
- W4386324079 title "Design of Image Recognition and Generation Algorithm Based on Artificial Neural Network" @default.
- W4386324079 cites W3010665394 @default.
- W4386324079 cites W3165843118 @default.
- W4386324079 doi "https://doi.org/10.1109/icdiime59043.2023.00028" @default.
- W4386324079 hasPublicationYear "2023" @default.
- W4386324079 type Work @default.
- W4386324079 citedByCount "0" @default.
- W4386324079 crossrefType "proceedings-article" @default.
- W4386324079 hasAuthorship W4386324079A5021355100 @default.
- W4386324079 hasConcept C11413529 @default.
- W4386324079 hasConcept C115961682 @default.
- W4386324079 hasConcept C134306372 @default.
- W4386324079 hasConcept C138885662 @default.
- W4386324079 hasConcept C153180895 @default.
- W4386324079 hasConcept C154945302 @default.
- W4386324079 hasConcept C175202392 @default.
- W4386324079 hasConcept C177148314 @default.
- W4386324079 hasConcept C2776401178 @default.
- W4386324079 hasConcept C33923547 @default.
- W4386324079 hasConcept C41008148 @default.
- W4386324079 hasConcept C41895202 @default.
- W4386324079 hasConcept C50644808 @default.
- W4386324079 hasConcept C57273362 @default.
- W4386324079 hasConceptScore W4386324079C11413529 @default.
- W4386324079 hasConceptScore W4386324079C115961682 @default.
- W4386324079 hasConceptScore W4386324079C134306372 @default.
- W4386324079 hasConceptScore W4386324079C138885662 @default.
- W4386324079 hasConceptScore W4386324079C153180895 @default.
- W4386324079 hasConceptScore W4386324079C154945302 @default.
- W4386324079 hasConceptScore W4386324079C175202392 @default.
- W4386324079 hasConceptScore W4386324079C177148314 @default.
- W4386324079 hasConceptScore W4386324079C2776401178 @default.
- W4386324079 hasConceptScore W4386324079C33923547 @default.
- W4386324079 hasConceptScore W4386324079C41008148 @default.
- W4386324079 hasConceptScore W4386324079C41895202 @default.
- W4386324079 hasConceptScore W4386324079C50644808 @default.
- W4386324079 hasConceptScore W4386324079C57273362 @default.
- W4386324079 hasLocation W43863240791 @default.
- W4386324079 hasOpenAccess W4386324079 @default.
- W4386324079 hasPrimaryLocation W43863240791 @default.
- W4386324079 hasRelatedWork W1494505385 @default.
- W4386324079 hasRelatedWork W1591671037 @default.
- W4386324079 hasRelatedWork W2007567623 @default.
- W4386324079 hasRelatedWork W2094559136 @default.
- W4386324079 hasRelatedWork W2382607599 @default.
- W4386324079 hasRelatedWork W2386387936 @default.
- W4386324079 hasRelatedWork W2546942002 @default.
- W4386324079 hasRelatedWork W2970216048 @default.
- W4386324079 hasRelatedWork W3132254705 @default.
- W4386324079 hasRelatedWork W4236038968 @default.
- W4386324079 isParatext "false" @default.
- W4386324079 isRetracted "false" @default.
- W4386324079 workType "article" @default.