Upload latest models, conv vs, swin

app.py

@@ -10,21 +10,34 @@ import gradio as gr
 # def is_cat(x): return x[0].isupper()
 
 # %% app.ipynb 3
-learn = load_learner("model_swin.pkl")
+learn_swin = load_learner("model_swin.pkl")
+learn_conv = load_learner("model_conv.pkl")
 
 # %% app.ipynb 5
-categories = ['Bathroom', 'Bedroom', 'Floor plan', 'Front', 'Kitchen', 'Living room', 'Parking', 'Porch', 'Swimming pool', 'Views', 'Yard']
-
-def classify_img(img):
-    pred,idx,probs = learn.predict(img)
+categories = [
+    'Bathroom', 'Bedroom', 'Floor plan', 'Front', 'Home Office', 'Kitchen', 
+    'Laundry', 'Living room', 'Parking', 'Porch', 'Swimming pool', 'Views', 
+    'Walk In Closet', 'Yard'
+]
+
+def classify_img(img, use_conv):
+    pred,idx,probs = learn_conv.predict(img) if use_conv else learn_swin.predict(img)
     return dict(zip(categories, map(float, probs)))
 
 
 # %% app.ipynb 7
-image = gr.inputs.Image(shape=(192, 192))
+image = gr.inputs.Image(shape=(240, 320))
 label = gr.outputs.Label()
 examples = ["kitchen.jpg", "living_room.jpg", "living_room2.jpg"]
 
-intf = gr.Interface(fn=classify_img, inputs=image, outputs=label, examples=examples)
+intf = gr.Interface(
+    fn=classify_img, 
+    inputs=[
+        gr.components.Image(type="pil", shape=(640, 480)), 
+        gr.components.Checkbox(label="Use conv model", value=False), 
+    ], 
+    outputs=label, 
+    examples=examples
+)
 intf.launch(inline=False)
 

model.pkl → model_conv.pkl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b3fdff1c13aeac79a260ff457afa208699fcb823ca30aa039b5d03f513ba03cb
-size 791545705
+oid sha256:ebe165e9dfd2333810ae84c91e84dcce5ea0522cc9dae1c7c1155a2a4a0835e8
+size 791581993

model_swin.pkl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7fc6a220984556aac5830721651d32b47f721f41dfe9891c2002cbe87740b737
-size 363515517
+oid sha256:d3558ca78c5c2ca865380f05c99cc777081a5eb765cd8ab286d6a74e4dd981cd
+size 797927745