Image datasets in design

Visual communication is an essential tool for designers, and the digital image has an important role both as a method of communication and as a method for capturing and storing information. In particular, the development of effective automated image collection methods has created new opportunities for designers to gather and analyze visual information about the places where they work. Trail cameras offer a simple and inexpensive way to collect large quantities of images, creating datasets that can reveal patterns and activities that might otherwise go unnoticed. These image datasets can serve as both qualitative and quantitative sources of information, offering different types of insights depending on how they are approached and analyzed.

A single trail camera image can yield multiple layers of information relevant to design practice. At the most basic level, an image documents the presence of particular species at a specific time and location, providing evidence of which animals use a site and when they are most active. With greater attention an image can reveal details like the activity an animal is engaged in, signs of health or sickness, ease or wariness. The same image may also contain detailed information about weather, characteristics of understory and tree canopy, and the spatial relationships between landscape elements. When viewed as part of a larger dataset, individual images contribute to understanding broader patterns: migration routes, territorial boundaries, seasonal behaviors, and the ways animals adapt to human-modified environments. This richness of information makes image datasets particularly valuable for designers who seek to understand the complex ecological and social dynamics of the sites where they propose interventions.

While general knowledge about animal behavior provides important background information, the particular conditions of each site create unique patterns of animal activity that can only be understood through direct observation. Setting up even a single trail camera on a potential design site can quickly reveal which animals are present, how they move through the space, and how their activities change throughout the day. The longer cameras can remain in place, the more comprehensive this understanding becomes, but even short-term deployments of a few days or weeks can provide valuable insights that inform more responsive and ecologically aware design decisions.

A trail camera will typically capture a modest number of images each day, but over the course of several days or a week it’s certainly possible to accumulate 1000 images or more, particularly if the trail camera sensitivity is too high. Machine learning becomes invaluable as the quantity of images grows, either through longer deployments or the use of multiple cameras. The uses of various forms of Artificial Intelligence (AI) with images is a complex topic that is developing quickly, and the methods mentioned in this book are simply the ones that we found helpful over the course of this project. We sincerely hope that some of these methods will also be relevant and useful for you.