Update app.py

app.py

@@ -57,7 +57,7 @@ LABELS_MAP = ["Bat (baseball)", "Bat (mammal)",
 
 """
 
-def run_classifer(model_key, image_path):
+def run_classifer(model_key, image_path, prob_threshold):
     # model_key: name of backbone zero-shot-image-classification model to use
     # image_path: path to test image
     # prob_threshold: confidence (i.e., probability) threshold above which to consider a prediction valid
@@ -87,8 +87,8 @@ def run_classifer(model_key, image_path):
     # Dictionary mapping all candidate labels to their predicted probabilities
     prob_dict    = {label_lookup[output['label']]: round(output["score"], 4) for output in outputs}
 
-    # predicted_label_str = f"This image shows {outputs[0]['label']} | Confidence (probability): {100*outputs[0]['score']:.1f}%" if outputs[0]['score'] > prob_threshold else "No prediction"
-    predicted_label_str = f"This image shows {outputs[0]['label']} | Confidence (probability): {100*outputs[0]['score']:.1f}%" 
+    predicted_label_str = f"This image shows {outputs[0]['label']} | Confidence (probability): {100*outputs[0]['score']:.1f}%" if float(outputs[0]['score']) > prob_threshold else "No prediction"
+    # predicted_label_str = f"This image shows {outputs[0]['label']} | Confidence (probability): {100*outputs[0]['score']:.1f}%" 
 
     return predicted_label_str, prob_dict
 
@@ -111,7 +111,7 @@ gradio_app = gradio.Interface(
     fn     = run_classifer,
     inputs = [gradio.Dropdown(["CLIP-base", "CLIP-large", "SigLIP-base", "SigLIP-large"], value="CLIP-large", label  = "Select Classifier"),
               gradio.Image(type="pil", label="Load sample image here"),
-              # gradio.Slider(minimum = 0.1, maximum = 0.9, step = 0.05, value = 0.25, label = "Set Prediction Threshold")
+              gradio.Slider(minimum = 0.1, maximum = 0.9, step = 0.05, value = 0.25, label = "Set Prediction Threshold")
               ],
 
     outputs = [gradio.Textbox(label="Image Classification"),