Scene Classification Using Deep Learning: A Critical Review of Representations, Benchmarks, and Evaluation
This narrative review examines deep learning for static visual scene classification across indoor, outdoor, and aerial imagery and concludes that useful progress should be assessed through reproducible gains under matched protocols and credible transfer to new environments, rather than isolated accuracy values.