Urban Street Photography Set

1,200 candid urban street frames across 14 cities — varied light, weather and density, faces blurred.

A diverse street-photography set for style training and scene understanding: crowded markets, empty dawn avenues, rain, neon night and hard noon light across 14 cities on four continents. Shot for visual variety rather than any single look, with recognisable faces and plates blurred for privacy.

Dataset details

Supported tasks

• Style-transfer and photographic LoRA training
• Scene-classification and urban-context models
• Composition and colour reference

Dataset structure

Each item ships as a photograph plus a metadata sidecar (no GPS beyond city level):

{
  "id": "street_00734",
  "image": "street_00734.jpg",
  "city": "Lisbon",
  "time_of_day": "blue_hour",
  "weather": "clear",
  "scene": "tram_stop",
  "privacy": { "faces_blurred": true, "plates_blurred": true }
}

Samples

Motion sample

A short clip rendered with this listing — click to play the HLS stream.

Curation

Frames were selected from the creator's multi-year street archive to maximise diversity of light, weather and density, then privacy-processed with automated face/plate blurring and a manual review pass. Location is stored only at city granularity.

Personal and sensitive information

People appear incidentally in public space; recognisable faces and licence plates are blurred. No names, no fine-grained location. Not intended for person re-identification.

Considerations & limitations

• Face blurring removes fine facial detail — not for face-related tasks.
• City coverage is uneven; some cities contribute far more frames.
• Photographer's compositional taste introduces a stylistic bias.

License & provenance

Shot by the creator; privacy-processed before release. The sample archive (full_dataset.zip) is a free 50-item preview; the full set of 1,200 items unlocks with a purchase. Redistribution of the raw dataset is not permitted under either license.

Citation

@misc{nova-labs_00000010,
  title  = {Urban Street Photography Set},
  author = {Nova Labs (@nova-labs)},
  year   = {2026},
  note   = {TensorForge marketplace dataset, LST-00000010-0000-4000-8000-000000000010}
}