MMArt&ACM'18- Proceedings of the 2018 International Joint Workshop on Multimedia Artworks Analysis and Attractiveness Computing in Multimedia


SESSION: Keynote Address

  • Toshihiko Yamasaki
 

Digital Art by teamLab

  • Takeshi Yamada

teamLab speaks about the theme of 'Relationships Among People', one of teamLab's concepts which aims to explore a new relationship among people, and to make the presence of others a positive experience through digital art. Yamada will introduce such concept along with teamLab's works. teamLab was founded in 2001 as an interdisciplinary art collective whose expansive practice involves collaborations in the fields of art, technology and science, exploring the new relationships between people, and between people and the world in the information age. Artists, programmers, engineers, CG animators, mathematicians, architects, web and print graphic designers and editors form teamLab. Using technology to dissolve the boundaries between the physical and conceptual, and to propose new models of perception in the digital era, their work is immersive and interactive-focused on the themes of creativity, play, exploration, immersion, life, and fluidity. teamLab has been the subject of numerous exhibitions at venues worldwide, including the vast exhibition "teamLab : Au-delà des limites" currently on view at La Villette in Paris.

SESSION: Multimedia Artworks Analysis

  • Wei-Ta Chu
 

Doodle Master: A Doodle Beautification System Based on Auto-encoding Generative Adversarial Networks

  • Chien-Wen Chen
  • Wen-Cheng Chen
  • Min-Chun Hu

For those people without artistic talent, they can only draw rough or even awful doodles to express their ideas. We propose a doodle beautification system named Doodle Master, which can transfer a rough doodle to a plausible image and also keep the semantic concepts of the drawings. The Doodle Master applies the VAE/GAN model to decode and generate the beautified result from a constrained latent space. To achieve better performance for sketch data which is more like discrete distribution, a shared-weight method is proposed to improve the learnt features of the discriminator with the aid of the encoder. Furthermore, we design an interface for the user to draw with basic drawing tools and adjust the number of reconstruction times. The experiments show that the proposed Doodle Master system can successfully beautify the rough doodle or sketch in real-time.

Photo Selection for Family Album using Deep Neural Networks

  • Sijie Shen
  • Toshihiko Yamasaki
  • Michi Sato
  • Kenji Kajiwara

The development of digital cameras and the web booming are the critical reasons for the increasing of digital portraits. However, such kind of daily photos are usually too many to select and organize, which leads to further requirement of better photo management services. In this paper, we are focusing on a significant part of daily photos -- family photos. We collaborate with a family photo service provider, Chikaku Inc., to create a family photo dataset. The dataset contains 12,140 images with corresponding rates (from 1 to 5) measuring if it is suitable to be selected for a family album. According to our experiments on classifying eligibility of the photos as a printed family photo album, the classification accuracy reaches 96.6% on the test set.

Predicting Image Aesthetics using Objects in the Scene

  • HIYA ROY
  • Toshihiko Yamasaki
  • Tatsuaki Hashimoto

Analyzing aesthetic quality of images is a highly challenging task because of its subjectiveness. With the exponential rise of digital images in social media, it is of great demand to assess the aesthetics of images for several multimedia applications such as increasing social popularity etc. Previous approaches to address this problem have used hand-designed features or automated features extracted by deep convolutional neural network architectures. In this paper we predict the aesthetics of images by using the inferential information depending on the visual content found in an image. To the best of our knowledge, this is the first attempt to address such problem by using tags predicted. Experimental results show that our proposed method outperforms the traditional machine learning methods and demonstrate competitive performance compared to the state-of-the-art methods of image aesthetics prediction.

SESSION: Attractiveness Computing in Multimedia

  • Norimichi Tsumura
 

Colour Perception Characteristics of Women in Menopause

  • Mayuko Iriguchi
  • Hiroki Koda
  • Nobuo Masataka

Female hormones affect perception, cognition and mental condition of women in the menstrual cycle and menopause. However, how menopause affects colour perception, and whether mental condition, especially depression, is related to colour perception remain unclear. Here, we investigated the influences of menopause on colour perception, recording the response times to three types of face and scrambled face stimuli: happy, neutral and sad, with three colours: red, yellow and blue. We predicted that colour perception of participants would be interfered with and delayed by emotional facial expression, as emotional facial expressions are connected with specific colours. We also examined depression states in women using the Center for Epidemiologic Studies Depression Scale (CES-D Scale) to understand the relationship between colour perception and depression. Fifty-nine female participants participated in the experiments. We analysed the data of 23 pre-menopausal and 20 post-menopausal participants. The results showed that the pre-menopausal women reacted to all stimuli faster than the post-menopausal women, and the post-menopausal women reacted to only blue significantly more slowly than to the other colours. Colour perception had no clear association with reaction time to emotional facial expression or with depression. The results suggest that menopause could influence colour perception of women as one of the possible factors, as well as aging, with such influence, and imply that the perceptual differences of blue between pre- and post-menopausal women result from a deficit in short wavelength sensitivity cones possibly caused by menopause-associated hormonal changes.

Temporal Course of Neural Processing during Skin Color Perception: An Event-related Potential Study

  • Hirokazu Doi
  • Norimichi Tsumura
  • Kazuyuki Shinohara

Recently, increasing number of behavioral studies have shown that subjective evaluation of health and attractiveness is modulated by facial skin color and pigment distribution. These studies, however, do not go further than describing the behavioral phenomena, and little is known about the neural mechanism underlying subjective evaluation of facial skin color. The present study investigated the temporal course of neural activation related to subjective evaluation of skin color of female faces. To achieve this goal, we analyzed the event-related potentials recorded while participants were evaluating attractiveness and perceived health of synthesized female faces with their skin color being systematically modulated. Principal component analysis revealed that perceived health modulated ERP at multiple time windows, while no prominent effect of perceived attractiveness was observed. This finding indicates the possibility that the visual system treats the facial skin color primarily as a clue of con-specific's health states. The implementation of such machinery is supposedly beneficial in mate-selection by increasing the odds of successful reproduction.

Modeling the Relation between Skin Attractiveness and Physical Characteristics

  • Kensuke Tobitani
  • Tatsuya Matsumoto
  • Yusuke Tani
  • Noriko Nagata

There is a wide range of requirements for representing skin quality in various fields, such as computer graphics and cosmetics. However, accurate representation of skin quality is a difficult technical issue, because of the complex physical characteristics of skin and its constituent substances and structure. The objective of this study is to clarify the latent factors in the impression of skin, mainly skin attractiveness and to provide systematic modeling of the relation between impression and physical characteristics. This will not only enable intuitive representation of skin quality, but will also help elucidate the human recognition mechanism in relation to skin. To build the model, we first select words from those that are generally used to assess skin. Next, we study techniques capable of accurate skin representation, based on the physical characteristics of skin, and create CG images of skin. We then use the words and images to clarify the latent factors for forming a visual impression of skin. Finally, we conduct a regression analysis on the results obtained in the subjective evaluation test to build a systematic model to estimate the impression evoked, based on the skin's physical characteristics. This makes it possible to estimate impressions from physical characteristics and conversely, to estimate those specific physical characteristics that contribute to a desired impression.