Question

Consider the doodles you found in the Quick, Draw! dataset:

What did you notice as you explored drawings in different categories? Are there consistent themes among the pictures in a category?

If you didn’t know the category labels, how would you distinguish the pictures from one another? What would you look for?

Reflect on your choices and think about categories with the pictures you just explored.

Now, write 2-3 sentences (40-60 words) in response to each of these questions. Type your response in the text box below.

Answer 1

Exploring doodles across categories revealed consistent motifs and simplified, exaggerated shapes—people often draw the defining features of an object (e.g., wings for birds, wheels for cars, eyes and beaks for faces) while omitting detail. Although line quality and style vary, the recurring elemental cues create clear visual signatures for each category.

Without labels I would rely on salient structural cues—overall silhouette, key components like wheels, stems, or beaks, relative proportions, and distinctive marks such as eyes, handles, or spikes—to cluster drawings by likely object. I would also consider stroke density and placement patterns, since habitual drawing sequences often reveal canonical parts.

Reflecting on my choices, I noticed that prototypical exemplars make categories easy to identify, while ambiguous or minimalist sketches blur boundaries and cause misclassification. This suggests category labels capture a shared mental image but dataset variability and cultural drawing habits require flexible, feature-based approaches for robust recognition.