Image Analysis
Image Scrape
To gather a comparative dataset of culturally available imagery, I used SerpAPI (2025), a third-party search tool that enables programmatic access to Google Image results. This approach was chosen over direct use of the Google Cloud Vision API (2025) due to significant restrictions on search query specificity and filtering imposed by Google. SerpAPI allowed for fine-tuned search queries that better aligned with the aims of this research: to analyse how culturally available images visually encode both evolved human needs and conditions of suppression or disconnection.
Search Criteria
To understand what humans really need to thrive, I examined how those needs and their opposites show up in everyday image culture. I searched two sets of terms: one based on evolved instincts for wellbeing, and one reflecting the modern forces that compete with them.
|
Connection Creativity Education Freedom Happiness Human Instinct Instinct Nature Play Safety |
Artificial Conformity Danger Disconnection Doomscrolling Isolation Sadness Social Media Addiction Standardisation Technology Addiction |
Search Modifiers
All search terms used a standardised negative search filter to exclude irrelevant, entertainment-driven content:
-movie -tv -series -logo -lyrics -song -album -music -watch -film -poster -text -quote -book -cover -title
Reflexivity and Bias
While this filtering strategy increased the relevance of the image set to the project’s thematic scope, it also introduced methodological bias. By predetermining which media types were excluded, I inevitably shaped the representational landscape. This may have filtered out alternative cultural expressions such as artistic, cinematic, or poetic interpretations of terms like ‘freedom’ or ‘play’ which could hold symbolic or affective power.
This approach was not neutral. It was guided by my Capstone’s specific aim: to foreground imagery that supports an evolutionary, ecological, and embodied understanding of what humans need to thrive. Nevertheless, I remained critically aware of the exclusions imposed by my own framing and acknowledge that this may have reinforced a particular worldview. In future research, a comparative method, juxtaposing filtered and unfiltered results, could provide richer insight into both dominant and marginal visual narratives.
Image Scraping Summary
-
Tool: SerpAPI
-
Searches conducted: 20 terms across two thematic groups
-
Images collected: ~ 300 – 500
-
Filters applied: -movie -tv -logo -quote -text etc.
-
Storage: Images were downloaded programmatically via the SerpAPI response and saved into keyword-specific folders for later visual analysis using Clarifai’s object recognition tools.
-
Use basis: Educational, non-commercial, under UK fair dealing