Comparar métodos

	Clasificación de Imágenes Multimodales	Clasificación de Imágenes Mediante Ajuste Fino
Campo	Aprendizaje profundo	Aprendizaje profundo
Familia	Machine learning	Machine learning
Año de origen	2011–2021	2010–2014
Autor original	Ngiam et al.; Radford et al. (CLIP)	Yosinski, J. et al.; Pan, S. J. & Yang, Q.
Tipo	Multimodal supervised classification	Transfer learning / fine-tuning
Fuente seminal	Radford, A., Kim, J. W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., ... & Sutskever, I. (2021). Learning transferable visual models from natural language supervision. Proceedings of the 38th International Conference on Machine Learning (ICML), PMLR 139, 8748–8763.	Yosinski, J., Clune, J., Bengio, Y., & Lipson, H. (2014). How transferable are features in deep neural networks? Advances in Neural Information Processing Systems (NeurIPS), 27, 3320–3328.
Alias	multimodal visual classification, image-text classification, vision-language classification, cross-modal image classification	fine-tuning for image recognition, transfer learning image classifier, pretrained CNN fine-tuning, domain-specific image classifier
Relacionados	6	5
Resumen	Multimodal image classification extends standard visual classification by incorporating additional modalities — such as text captions, audio, or structured metadata — alongside image features. Separate encoders process each modality, their representations are fused, and a joint classifier assigns the target label. Models such as CLIP demonstrate that image–text alignment enables zero-shot and few-shot image classification at scale.	Fine-tuned image classification adapts a large neural network pretrained on a broad image corpus (such as ImageNet) to a specific target domain by continuing training on labeled domain images. This approach achieves strong accuracy with far fewer target-domain samples than training from scratch, making it the dominant paradigm for applied computer vision tasks.