As facial color pattern around the eyes has been suggested to serve various adaptive functions related to the gaze signal, we compared the patterns among 25 canid species, focusing on the gaze signal, to estimate the function of facial color pattern in these species. The facial color patterns of the studied species could be categorized into the following three types based on contrast indices relating to the gaze signal: A-type (both pupil position in the eye outline and eye position in the face are clear), B-type (only the eye position is clear), and C-type (both the pupil and eye position are unclear). A-type faces with light-colored irises were observed in most studied species of the wolf-like clade and some of the red fox-like clade. A-type faces tended to be observed in species living in family groups all year-round, whereas B-type faces tended to be seen in solo/pair-living species. The duration of gazing behavior during which the facial gaze-signal is displayed to the other individual was longest in gray wolves with typical A-type faces, of intermediate length in fennec foxes with typical B-type faces, and shortest in bush dogs with typical C-type faces. These results suggest that the facial color pattern of canid species is related to their gaze communication and that canids with A-type faces, especially gray wolves, use the gaze signal in conspecific communication.

For deep learning applications, the massive data development (e.g., collecting, labeling), which is an essential process in building practical applications, still incurs seriously high costs. In this work, we propose an effective data augmentation method based on generative adversarial networks (GANs), called Domain Fusion. Our key idea is to import the knowledge contained in an outer dataset to a target model by using a multi-domain learning GAN. The multi-domain learning GAN simultaneously learns the outer and target dataset and generates new samples for the target tasks. The simultaneous learning process makes GANs generate the target samples with high fidelity and variety. As a result, we can obtain accurate models for the target tasks by using these generated samples even if we only have an extremely low volume target dataset. We experimentally evaluate the advantages of Domain Fusion in image classification tasks on 3 target datasets: CIFAR-100, FGVC-Aircraft, and Indoor Scene Recognition. When trained on each target dataset reduced the samples to 5,000 images, Domain Fusion achieves better classification accuracy than the data augmentation using fine-tuned GANs. Furthermore, we show that Domain Fusion improves the quality of generated samples, and the improvements can contribute to higher accuracy.

The rotation prediction (Rotation) is a simple pretext-task for selfsupervised learning (SSL), where models learn useful representations for target vision tasks by solving pretext-tasks. Although Rotation captures information of object shapes, it hardly captures information of textures. To tackle this problem, we introduce a novel pretext-task called image enhanced rotation prediction (IE-Rot) for SSL. IE-Rot simultaneously solves Rotation and another pretext-task based on image enhancement (e.g., sharpening and solarizing) while maintaining simplicity. Through the simultaneous prediction of rotation and image enhancement, models learn representations to capture the information of not only object shapes but also textures. Our experimental results show that IE-Rot models outperform Rotation on various standard benchmarks including ImageNet classification, PASCAL-VOC detection, and COCO detection/segmentation.

The accuracy of deep neural networks is degraded when the distribution of features in the test environment (target domain) differs from that of the training (source) environment. To mitigate the degradation, test-time adaptation (TTA), where a model adapts to the target domain without access to the source dataset, can be used in the test environment. However, the existing TTA methods lack feature distribution alignment between the source and target domains, which unsupervised domain adaptation mainly addresses, because accessing the source dataset is prohibited in the TTA setting. In this paper, we propose a novel TTA method, named Covariance-Aware Feature alignment (CAFe), which explicitly aligns the source and target feature distributions at test time. To perform alignment without accessing the source data, CAFe uses auxiliary feature statistics (mean and covariance) pre-computed on the source domain, which are lightweight and easily prepared. Further, to improve efficiency and stability, we propose feature grouping, which splits the feature dimensions into groups according to their correlations by using spectral clustering to avoid degeneration of the covariance matrix. We empirically show that CAFe outperforms prior TTA methods on a variety of distribution shifts.Preprint. Under review.

No abstract