6,019
Views
9
CrossRef citations to date
0
Altmetric
Research Article

Automated Visual Content Analysis (AVCA) in Communication Research: A Protocol for Large Scale Image Classification with Pre-Trained Computer Vision Models

ORCID Icon, ORCID Icon &

Figures & data

Figure 1. A research protocol of using computer vision APIs for communication research

Figure 1. A research protocol of using computer vision APIs for communication research

Table 1. Metrics for the best-performing model per method: validation set

Figure 2. Results for profit

Dotted line shows the frequency of the most common category, and it would be expected models performing well would at least surpass this threshold.
Figure 2. Results for profit

Figure 3. Results for people

Figure 3. Results for people

Figure 4. Results for planet

Figure 4. Results for planet

Table 2. Metrics for the overall best-performing model: test set

Table 3. Frequency of images in the full sample per sustainability dimension (N = 21,841)

Table 4. Application contexts for AVCA across communication studies