Update app.py

app.py

@@ -10,7 +10,6 @@ from PIL import Image
 labels = ["drawings", "hentai", "neutral", "porn", "sexy"]
 description = f"""This is a demo of classifing nsfw pictures. Label division is based on the following:
 [*https://github.com/alex000kim/nsfw_data_scraper*](https://github.com/alex000kim/nsfw_data_scraper).
-(If you want to test, please drop the example pictures instead of clicking)
 
 You can continue to train this model with the same preprocess-to-images.
 Finally, welcome to star my [*github repository*](https://github.com/csuer411/nsfw_classify)""" 
@@ -49,17 +48,7 @@ model.load_state_dict(torch.load("classify_nsfw_v3.0.pth", map_location="cpu"))
 model.eval()
 
 
-def img_convert(inp):
-    with io.BytesIO() as f:
-        inp.save(f, format="JPEG")
-        img_data = f.getvalue()
-    img_base64 = base64.b64encode(img_data)
-    return img_base64
-
-
-
 def predict(inp):
-    temp_inp = inp
     inp = preprocess(inp).unsqueeze(0)
     with torch.no_grad():
         prediction = torch.nn.functional.softmax(model(inp)[0], dim=0)
@@ -67,8 +56,8 @@ def predict(inp):
     return result
 
 
-inputs = gr.components.Image(type='pil')
-outputs = gr.components.Label(num_top_classes=2)
+inputs = gr.Image(type='pil')
+outputs = gr.Label(num_top_classes=2)
 gr.Interface(
     fn=predict, inputs=inputs, outputs=outputs, examples=["./example/anime.jpg", "./example/real.jpg"], description=description,
 ).launch()