first commit

app.py

@@ -0,0 +1,59 @@
+import gradio as gr
+import re
+from transformers import DonutProcessor, VisionEncoderDecoderModel
+import torch
+from PIL import Image
+
+def process_filename(filename, question):
+    print(f"Image file: {filename}")
+    print(f"Question: {question}")
+    image = Image.open(filename).convert("RGB")
+    return process_image(image)
+
+
+def process_image(image, question):
+    processor = DonutProcessor.from_pretrained("naver-clova-ix/donut-base-finetuned-docvqa")
+    model = VisionEncoderDecoderModel.from_pretrained("naver-clova-ix/donut-base-finetuned-docvqa")
+
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+    model.to(device)
+
+    # prepare decoder inputs
+    prompt = f"<s_docvqa><s_question>{question}</s_question><s_answer>"
+    decoder_input_ids = processor.tokenizer(prompt, add_special_tokens=False, return_tensors="pt").input_ids
+
+    pixel_values = processor(image, return_tensors="pt").pixel_values
+
+    outputs = model.generate(
+        pixel_values.to(device),
+        decoder_input_ids=decoder_input_ids.to(device),
+        max_length=model.decoder.config.max_position_embeddings,
+        pad_token_id=processor.tokenizer.pad_token_id,
+        eos_token_id=processor.tokenizer.eos_token_id,
+        use_cache=False,
+        bad_words_ids=[[processor.tokenizer.unk_token_id]],
+        return_dict_in_generate=True,
+    )
+
+    sequence = processor.batch_decode(outputs.sequences)[0]
+    sequence = sequence.replace(processor.tokenizer.eos_token, "").replace(processor.tokenizer.pad_token, "")
+    sequence = re.sub(r"<.*?>", "", sequence, count=1).strip()  # remove first task start token
+    print(processor.token2json(sequence))
+
+    return [True, processor.token2json(sequence)['answer'], ""]
+
+def process_document(image, question):
+    ret = process_image(image, question)
+    return ret[1]
+
+description = "DocVQA (document visual question answering)"
+
+demo = gr.Interface(
+    fn=process_document,
+    inputs=["image", gr.Textbox(label = "Question" )],
+    outputs=gr.Textbox(label = "Response" ),
+    title="Extract data from image",
+    description=description,
+    cache_examples=True)
+
+demo.launch()

requirements.txt

@@ -0,0 +1,3 @@
+gradio
+transformers
+torch