Updated from colab

app.py

@@ -0,0 +1,17 @@
+from transformers import AutoModelForImageClassification, AutoFeatureExtractor
+import gradio as gr
+import torch
+
+huggingface_username = 'i-am-holmes'
+model_name = 'vit-base-patch16-224-finetuned-flower'
+
+def classify_image(image):
+  device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
+  model = AutoModelForImageClassification.from_pretrained(f'{huggingface_username}/{model_name}').to(device)
+  feature_extractor = AutoFeatureExtractor.from_pretrained(f'{huggingface_username}/{model_name}')  
+  inp = feature_extractor(image, return_tensors='pt').to(device)
+  outp = model(**inp)
+  pred = torch.argmax(outp.logits, dim=1).item()
+  return model.config.id2label[pred]
+
+interface = gr.Interface(fn=classify_image, inputs=gr.Image(shape=(224, 224)), outputs="text").launch(debug=True, share=True)

requirements.txt

@@ -0,0 +1,2 @@
+torch
+transformers