Search

This paper presents an end-to-end attention-based neural network for identifying writers in historical documents. The proposed network does not require any preprocessing stages such as binarization or segmentation. It has three main parts: a feature extractor using Convolutional Neural Network (CNN) to extract features from an input image; an attention filter to select key points; and a generalized deep neural VLAD model to form a representative vector by aggregating the extracted key points. The whole network is trained end-to-end by a combination of cross-entropy and triplet losses. In the experiments, we evaluate the performance of our model on the HisFragIR20 dataset that consists of about 120,000 historical fragments from many writers. The experiments demonstrate better mean average precision and accuracy at top-1 in comparison with the state-of-the-art results on the HisFragIR20 dataset. This model is rather new for dealing with various sizes of historical document fragments in the writer identification and image retrieval.

In this paper, we propose an RNN-Transducer model for recognizing Japanese and Chinese offline handwritten text line images. As far as we know, it is the first approach that adopts the RNN-Transducer model for offline handwritten text recognition. The proposed model consists of three main components: a visual feature encoder that extracts visual features from an input image by CNN and then encodes the visual features by BLSTM; a linguistic context encoder that extracts and encodes linguistic features from the input image by embedded layers and LSTM; and a joint decoder that combines and then decodes the visual features and the linguistic features into the final label sequence by fully connected and softmax layers. The proposed model takes advantage of both visual and linguistic information from the input image. In the experiments, we evaluated the performance of the proposed model on the two datasets: Kuzushiji and SCUT-EPT. Experimental results show that the proposed model achieves state-of-the-art performance on all datasets.

This paper presents an approach for matching various sizes of excavated wooden fragments based on siamese neural networks. We propose a siamese network composed of global average pooling and a fine-tuned Resnet encoder (GA-S-net). We also propose an elaborated siamese network by replacing the global average pooling with a spatial pyramid pooling layer and add a new dense absolute difference layer (SP-S-net). Samples of 37,760 fragments were prepared from 268 complete wooden tablets excavated from the Heijo-Kyo Palace ruins used during the Nara period in Japan. Both of the networks answer whether two fragments are from the same tablet or not. The result of both networks for the testing set is similar to AUC (Area under the curve) of ROC (Receiver Operating Characteristic) curve being around 90%. In AUC of large fragments, however, SP-S-net is better than GA-S-net with 97.1% versus 93.8%. These networks are rather new for dealing with various sizes of inputs for the matching problem.