Update app.py

app.py

@@ -1,76 +1,34 @@
+import os
 import gradio as gr
 from transformers import pipeline
 
-pipe2 = pipeline("automatic-speech-recognition", model="distil-whisper/distil-small.en")
-pipe3 = pipeline("automatic-speech-recognition", model="antony66/whisper-large-v3-russian")
+def launch(input_image):
+    out = depth_estimator(input_image)
 
-demo = gr.Blocks()
-
-
-def transcribe_speech_english(filepath):
-    if filepath is None:
-        gr.Warning("No audio found, please retry.")
-        return ""
-    output = pipe2(filepath)
-    return output["text"]
-
-
-def transcribe_speech_russian(filepath):
-    if filepath is None:
-        gr.Warning("No audio found, please retry.")
-        return ""
-    output = pipe3(filepath)
-    return output["text"]
-
-
-mic_transcribe_english = gr.Interface(
-    fn=transcribe_speech_english,
-    inputs=gr.Audio(sources="microphone",
-                    type="filepath"),
-    outputs=gr.Textbox(label="Transcription",
-                       lines=3),
-    allow_flagging="never")
-
-
-mic_transcribe_russian = gr.Interface(
-    fn=transcribe_speech_russian,
-    inputs=gr.Audio(sources="microphone",
-                    type="filepath"),
-    outputs=gr.Textbox(label="Transcription",
-                       lines=3),
-    allow_flagging="never")
-
-
-file_transcribe_english = gr.Interface(
-    fn=transcribe_speech_english,
-    inputs=gr.Audio(sources="upload",
-                    type="filepath"),
-    outputs=gr.Textbox(label="Transcription",
-                       lines=3),
-    allow_flagging="never",
-)
+    # resize the prediction
+    prediction = F.interpolate(
+        out["predicted_depth"].unsqueeze(1),
+        size=input_image.size[::-1],
+        mode="bicubic",
+        align_corners=False,
+    )
 
+    # normalize the prediction
+    output = prediction.squeeze().numpy()
+    formatted = (output * 255 / np.max(output)).astype("uint8")
+    depth = Image.fromarray(formatted)
+    return depth
 
-file_transcribe_russian = gr.Interface(
-    fn=transcribe_speech_russian,
-    inputs=gr.Audio(sources="upload",
-                    type="filepath"),
-    outputs=gr.Textbox(label="Transcription",
-                       lines=3),
-    allow_flagging="never",
-)
+iface = gr.Interface(launch,
+                     inputs=gr.Image(type='pil'),
+                     outputs=gr.Image(type='pil'))
 
+demo = gr.Blocks()
 
 with demo:
     gr.TabbedInterface(
-        [mic_transcribe_english,
-         file_transcribe_english,
-         mic_transcribe_russian,
-         file_transcribe_russian],
-        ["Transcribe Microphone English",
-         "Transcribe Audio File English",
-         "Transcribe Microphone Russian",
-         "Transcribe Audio File Russian"],
+        [iface],
+        ["iface"],
     )
 
-demo.launch()
+demo.launch(debug=True)