Animal Detection Dataset with Bounding Boxes for 200 Species

Horama/animal-200-detection is a subset of scraped wildlife images automatically annotated for object detection. It contains bounding boxes for approximately 200 animal species, generated using Grounding DINO and filtered for alive subjects without watermarks. The dataset was created by Horama and last updated on April 7, 2026.

Use Cases

• Train animal detection models based on the automatically generated bounding boxes.
• Benchmark object detection algorithms on a multi-species wildlife dataset.
• Fine-tune pre-trained vision models for ecological monitoring based on the wildlife image content.
• Develop educational tools for species identification based on the annotated visual data.

Strengths

• Contains bounding box annotations for approximately 200 animal species.
• Images were filtered for alive subjects and the absence of watermarks.
• Annotations were generated automatically using the Grounding DINO model.

Limitations

• Row count, column definitions, and file formats are unknown.
• Annotations are automatically generated, which may introduce noise compared to human-labeled data.
• The dataset's license is unspecified, which may restrict usage.

Provenance

Source

Horama/animal_scrapped (aka Horama/inat-raw), with images from DuckDuckGo, Wikimedia Commons, and Wikipedia.

Collection Method

Images were scraped, pre-classified with CLIP zero-shot, and annotated automatically with Grounding DINO.

Freshness

Last updated 2026-04-07 09:44:55.