Methodological Reflections on Ways of Seeing

Title of the Paper

Methodological Reflections on Ways of Seeing

Paper Information

• Field of Study: Application of Human-Computer Interaction (HCI) and Artificial Visual Intelligence in Creative Practices
• Keywords: Visual experience, meaning construction, artificial visual intelligence, creative practice, human-computer collaboration

Research Background and Problem Statement

• Identified Issues and Challenges: With the advancement of computational vision technologies, they have not only succeeded in practical applications (e.g., defect detection, weather forecasting) but also demonstrated new aesthetic forms in image generation (e.g., neural aesthetics and synthetic realities). However, the way computer vision processes visual information is often "black-boxed," and interdisciplinary understanding of its impact on human creative practices remains insufficient. The author points out that the collaborative methods between computer vision and human designers in creative practices have yet to be fully explored.
• Significance: The process of "seeing" for human designers involves more than observation; it encompasses creative meaning construction (e.g., visual thinking and image composition). Collaboration with computer vision could bring new possibilities to design creativity, but it requires further methodological exploration.
• Research Motivation and Related Work: All visual media, from photography and television to computers, have transformed the ways of seeing and producing images. While many studies have explored applications of computer vision in design, there is limited research on the cognitive process of "seeing" in designers' creative work, especially in collaboration with computer vision systems.

Solution

• Proposed Method or Solution: The author employs practice-oriented visual inquiry to compare how human designers and computer vision process visual information in creative practices.
• Innovative Aspects: The study conducts an in-depth analysis of the visual creation process from a first-person reflective perspective, exploring how designers discover meaning through the interaction of "seeing" and "doing" and collaborate with artificial visual intelligence for deep meaning construction.
• Implementation Steps and Techniques:
  • Through self-observation and analysis, document how designers construct meaning while working with photographic images;
  • Apply computer vision tools (e.g., IBM Watson Visual Recognition) to analyze how the same set of images is processed;
  • Compare the subjective composition methods of human designers with the specific labels and prediction scores returned by computer vision.

Research Outcomes

• Specific Findings:
  • The act of "seeing" is a dynamic, non-linear process of meaning construction, involving identifying, relating, and projecting visual and semantic relationships.
  • Human designers tend to use reflective judgment, while computer vision focuses on deterministic judgment (e.g., object recognition and color tagging).
  • Designers discover meaning through marking and abstract composition, while the labels returned by computer vision help guide designers to notice previously overlooked details.
• Advantages Compared to Existing Solutions:
  • Proposes a new model of artificial visual intelligence as a creative agent, emphasizing AI-guided assistance to help designers discover unnoticed details;
  • Combines the methodology with visual ideation, contributing to the expansion of design methods.
• Experimental or Evaluation Results:
  • Experiments based on visual composition results of specific photographic images reveal that the labels suggested by computer vision sometimes deviate from human designers' cognition but also provide novel perspectives due to their heterogeneity.
• Limitations and Future Directions:
  • The study pays limited attention to the technical details of computer vision algorithms;
  • Future work could develop systematic methods to link visual design with computational intelligence, supporting multi-user collaboration and personalized design.

Summary and Discussion

• Discussion Points:
  • Computational vision technology, as part of an "Agency of Noticing," can help designers complement their subjective perspectives;
  • Collaboration between humans and artificial intelligence should foster constructive interaction rather than simple automation.
  • Timing the intervention of AI in designers' workflows is crucial for inspiring creative ideas.
• Methodological and Practical Implications:
  • This study proposes a reflective and generative approach through visual abstraction and meaning construction, applicable to academic research and practical design.
  • Designers can leverage semantic tags from artificial visual intelligence to trigger new ideas while maintaining the subjectivity of the design process.

Directions for Future Research

• Explore how design education can transform computer vision into a participatory tool for visual design;
• Enhance the semantic-level training of artificial visual intelligence to support multidimensional and interdisciplinary creative practices.