Sample Anime Images With Model Predictions

Aggregating sample images generated by the TADNE model. It includes prediction results from an anime-face-detector (YOLOv3 + HRNetV2) and deepdanbooru tag predictions, including intermediate 4096-dimensional feature vectors.

Use Cases

• Fine-tune an anime face detection model using the YOLOv3 + HRNetV2 prediction results.
• Train a tag prediction model using the deepdanbooru outputs as labels for the generated images.
• Use the 4096-dimensional intermediate features from deepdanbooru for image similarity search or feature analysis.

Strengths

• Includes outputs from two specific models: an anime-face-detector and deepdanbooru.
• Provides intermediate 4096-dimensional feature vectors from the deepdanbooru model for deeper analysis.

Limitations

• The dataset size, number of images, and specific file formats are unknown.
• The data is a sample set, which may not be representative for training large-scale models.
• The last update was in April 2022, making the data potentially stale for current model comparisons.

Provenance

Source

hysts on Hugging Face

Collection Method

Images generated by the TADNE model, with predictions from specified external models.

Freshness

2022-04-15