README.md

metadata

license: mit
tags:
  - image-classification
  - pytorch
  - furry

Species Classification in Furry Art

This model was trained on ~16k images (filtered from ~40k). I may train it on a larger dataset in the future.

The model is capable of classifying about 86 species, including a few more common Pokemon, as well as some mythological and fictional species. The results may differ depending on the species, though the model is quite good with common ones such as foxes and wolves.

How to Use

Example using Pipelines (recommended):

from transformers import pipeline

classify = pipeline('image-classification', model='ashduino101/furry-species-classification')

scores = classify('your-image.png')
# Do something with the scores

Example using AutoModelForImageClassification:

import torch
from PIL import Image
from transformers import AutoImageProcessor, AutoModelForImageClassification

image_processor = AutoImageProcessor.from_pretrained('ashduino101/furry-species-classification')
inputs = image_processor(Image.open('your-image.png'), return_tensors='pt')

model = AutoModelForImageClassification.from_pretrained("ashduino101/furry-species-classification")
with torch.no_grad():
    result = model(**inputs)

# Do something with the result