Ground Truth (n.)
- The reality of a situation as experienced firsthand rather than by report
- Factual data as ascertainable through direct observation rather than through inference from remote sensing.
- In remote sensing, ground truth refers to information collected on location. Ground truth allows image data to be related to real features and materials on the ground.
- In statistics and machine learning, ground truth means checking the results of machine learning for accuracy against the real world.
Ground Truth (2021), by Sheung Yiu, combines documentary photography, archival imagery and abstract datasets to explore the relationship between human perception and optical technology. Set mainly in the forests of Finland, the project follows a group of scientists gathering ‘Ground Truth’ data to interpret satellite imagery.
In observational science, satellite imaging is frequently used to gather information on physical structures, measuring the electromagnetic radiation between the Earth and Sun. This can be instrumental in collecting data over large surface areas and can reveal details which were previously invisible, filling in the gaps of limited, earthbound perception. Rather than simply replacing manual observation, however, this technology creates a new role: the satellite data on its own is difficult to decipher and needs supplementing with additional information collected by scientists on the ground. As described by Sheung Yiu, the process is “boresome and borderline obsessive” and requires tasks such as weighing every individual leaf on a tree in order to calculate its total area. It’s often thought that technology would make this tireless labour redundant, but Ground Truth reveals the human efforts that are required to support the work of supposedly intelligent machinery. This symbiotic working relationship between people and technology is a central theme throughout.
It is a refreshingly optimistic and dignified account of how humans and machines can cooperate, at a time when fears about artificial intelligence stealing jobs and harvesting personal data seem to dominate the popular discourse.
The images in the book bring about an awareness of the co-dependency and alliance between humans and machines at a time when discourse about the dangers of artificial intelligence is more ubiquitous than ever. Spectral, abstract satellite imagery is presented alongside documentary photographs of the scientists who translate the data – and in doing so human and machine are cast as equals. This sequencing of images creates a sense of camaraderie and alliance between the two, an alliance which is apparent in the very nature of their dependent working relationship. It is a refreshingly optimistic and dignified account of how humans and machines can cooperate, at a time when fears about artificial intelligence stealing jobs and harvesting personal data seem to dominate the popular discourse. Sheung Yiu seems aware of this and photographs the machinery and equipment with the same respect as the people themselves. Shots of people and technology often share a double spread and are presented with equal status and significance on the page.
The codependency between humans and technology is furthered by the differences in their working processes. The book’s latter half (pages 113-166) features sample images produced by different types of technology scientists use on the ground. The brown pages are dense with grids of images from segment scans, photon simulations, hemispherical photographs and multiple other methods, all documenting and measuring the physical properties of the forest. The result is a mass of images, formatted algorithmically on the page, indicative of technologies’ ability to compile data at a huge rate, in a brute force style of collection. In the process of interpreting this information, the scientists on the ground must analyse and compare this data using specific, and more subjective methods. The process is slow and excruciating – something which is reflected in the photographs. The quiet, perceptive images often show the workers observing, measuring, setting up equipment and reviewing data obsessively, elaborating further on the differences in human/machine processes. The workflow on the ground is more measured and deliberate, whereas machine intelligence works with ruthless efficiency and speed, but lacks the thoughtfulness to analyse and interpret the data it collects.
The pacing of Ground Truth feels not unlike flicking through a scientist’s notebook; a window into the complex workflow of data collection.
Another interesting element of the datasets is how seductive they are in their abstraction. The average viewer would most likely not be able to understand the full significance of these models and is left instead with a set of visually satisfying graphic images. Likewise, scientists’ handwritten notes are sprawled across some pages, containing information which is meaningless to an average viewer, and instead functions to set the scene for the book, again highlighting the slower, analytical process of the scientists on the ground. This asks a larger question of where Ground Truth belongs. Photography has been in the joint custody of science, art and documentary for most of its existence – and by combining elements from all three, Sheung Yiu raises questions about photography’s role as a hybrid medium in contemporary visual culture, demonstrating how recontextualising the image can shift the way that it is interpreted by a viewer.
The book’s design is also a nod to this blending of genres. It feels as though it is intended to be consumed in a slow, thoughtful way. The documentary photographs often sit next to handwritten note pages, with scrawled measurements and notations, placing the viewer amongst these scientists at work. Often one page in a double spread is left blank, contributing to the quiet atmospheric working environment in the Finnish forests. The pacing of Ground Truth feels not unlike flicking through a scientist’s notebook; a window into the complex workflow of data collection.
Ground Truth, whilst not making anyone an expert, gives a unique insight into this esoteric field of science. It also constructs a narrative which sits outside the slipstream of technological fear and angst – giving a refreshing perspective on human/machine cooperation.