Privacy-Preserving Image Captioning with Deep Learning and Double Random Phase Encoding

Cloud storage has become eminent, with an increasing amount of data being produced daily; this led to substantial concerns related privacy and unauthorized access. To secure privacy, users can protect their private by uploading encrypted the cloud. Data encryption allows computations be performed on without decrypted in cloud, which requires enormous computation resources prevents access data. analysis such as classification, image query retrieval preserve if is using This paper proposes image-captioning method that generates captions over images encoder–decoder framework attention a double random phase encoding (DRPE) scheme. The are DRPE them then fed encoder adopts ResNet architectures generate fixed-length vector representations or features. decoder designed long short-term memory process features embeddings descriptive for images. We evaluate predicted BLEU, METEOR, ROUGE, CIDEr metrics. experimental results demonstrate feasibility our privacy-preserving captioning popular benchmark Flickr8k dataset.