Read without the caption
Describe the image's visible forms and relationships first. Note what remains uncertain. This provides a baseline for seeing what the text actually adds.
Read the caption closely
Separate factual identification from interpretation. Ask whether the text names a subject, supplies context, or proposes a meaning that the image alone cannot establish.
Compare the two readings
Write a brief account of what changed after reading the caption. Keep one observation that depends on the picture and one that depends on the text.