[Admin maintenance] Support new ZeroGPU hardware

README.md

@@ -4,7 +4,7 @@ emoji: 🏃
 colorFrom: purple
 colorTo: blue
 sdk: gradio
-sdk_version: 5.19.0
+sdk_version: 5.50.0
 app_file: app.py
 pinned: false
 license: mit

app.py

@@ -1,14 +1,33 @@
+import spaces
+import subprocess
+subprocess.run(["pip", "install", "--no-deps", "gradio-image-prompter"])
+
 import gradio as gr
 import numpy as np
 import torch
 from PIL import Image
 from gradio_image_prompter import ImagePrompter
+import gradio_image_prompter.image_prompter as _ip
+from gradio.data_classes import FileData
 from transformers import AutoProcessor, UdopForConditionalGeneration
 import easyocr
 from PIL import Image
-import spaces
 from typing import Optional, List, TypedDict, Union, Literal
 
+# gradio_image_prompter 0.1.0 was built for gradio 4.x; under gradio 5 its
+# postprocess returns an ImageData that no longer subclasses FileData, so the
+# PromptData model raises a pydantic ValidationError. Convert it back.
+_orig_img_post = gr.Image.postprocess
+def _patched_ip_postprocess(self, y):
+    if y is None:
+        return None
+    image, points = y.get("image", None), y.get("points", [])
+    img = _orig_img_post(self, image)
+    if img is not None and not isinstance(img, FileData):
+        img = FileData(**img.model_dump())
+    return _ip.PromptData(image=img, points=points)
+ImagePrompter.postprocess = _patched_ip_postprocess
+
 
 class PromptValue(TypedDict):
     image: Optional[Union[Image.Image, str]]
@@ -42,7 +61,7 @@ def udop_box_inference(image, text_prompt, box_coordinates):
     for box in bboxs:
       norm_boxes.append(normalize_bbox(box, width, height))
 
-    encoding = processor(image, text_prompt, texts, boxes=norm_boxes, return_tensors="pt")
+    encoding = processor(image, text_prompt, text_pair=texts, boxes=norm_boxes, return_tensors="pt")
     predicted_ids = model.generate(**encoding)
     return processor.batch_decode(predicted_ids, skip_special_tokens=True)[0]
 

requirements.txt

@@ -1,5 +1,4 @@
 git+https://github.com/huggingface/transformers.git
-gradio-image-prompter
 easyocr
 spaces
 torch