Upload 3 files

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+NotoSansSC-Regular.ttf filter=lfs diff=lfs merge=lfs -text

NotoSansSC-Regular.ttf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5cf8b2a0576d5680284ab03a7a8219499d59bbe981a79bb3dc0031f251c39736
+size 10560616

app.py

@@ -0,0 +1,89 @@
+import gradio as gr
+import pytesseract
+from PIL import Image, ImageDraw, ImageFont
+import requests
+import os
+import uuid
+from io import BytesIO
+from dotenv import load_dotenv
+
+load_dotenv()
+
+AZURE_KEY = os.getenv("AZURE_TRANSLATOR_KEY")
+AZURE_REGION = os.getenv("AZURE_TRANSLATOR_REGION")
+
+def translate_text_azure(text: str, to_lang="en"):
+    if not text.strip():
+        return ""
+    url = f"https://api.cognitive.microsofttranslator.com/translate?api-version=3.0&to={to_lang}"
+    headers = {
+        "Ocp-Apim-Subscription-Key": AZURE_KEY,
+        "Ocp-Apim-Subscription-Region": AZURE_REGION,
+        "Content-type": "application/json",
+    }
+    body = [{"text": text}]
+    response = requests.post(url, headers=headers, json=body)
+    return response.json()[0]["translations"][0]["text"]
+
+def extract_and_translate(image: Image.Image):
+    # OCR
+    ocr_results = pytesseract.image_to_data(image, output_type=pytesseract.Output.DICT)
+
+    texts = []
+    boxes = []
+    for i in range(len(ocr_results['text'])):
+        text = ocr_results['text'][i].strip()
+        if text:
+            (x, y, w, h) = (ocr_results['left'][i], ocr_results['top'][i],
+                            ocr_results['width'][i], ocr_results['height'][i])
+            translated = translate_text_azure(text)
+            texts.append({
+                "original": text,
+                "translated": translated,
+                "box": (x, y, w, h)
+            })
+            boxes.append((x, y, w, h))
+
+    return image, texts
+
+def render_translated_image(image, translations):
+    image = image.copy()
+    draw = ImageDraw.Draw(image)
+    font = ImageFont.load_default()
+    for t in translations:
+        x, y, w, h = t["box"]
+        draw.rectangle((x, y, x+w, y+h), fill="white")  # erase original text
+        draw.text((x, y), t["translated"], fill="black", font=font)
+    return image
+
+def pipeline(img):
+    base_img, translations = extract_and_translate(img)
+    table_data = [[t["original"], t["translated"]] for t in translations]
+    return base_img, table_data, translations
+
+def update_and_render(img, translations, updated_translations):
+    for i, updated in enumerate(updated_translations):
+        translations[i]["translated"] = updated[1]
+    return render_translated_image(img, translations)
+
+with gr.Blocks() as demo:
+    with gr.Row():
+        input_image = gr.Image(type="pil", label="Upload Manga Page")
+        output_image = gr.Image(type="pil", label="Translated Image")
+
+    table = gr.Dataframe(headers=["Original", "Translated"], interactive=True)
+    hidden_state = gr.State()
+
+    with gr.Row():
+        translate_btn = gr.Button("Extract & Translate")
+        apply_btn = gr.Button("Render Final Image")
+
+    translate_btn.click(fn=pipeline,
+                        inputs=input_image,
+                        outputs=[input_image, table, hidden_state])
+
+    apply_btn.click(fn=update_and_render,
+                    inputs=[input_image, hidden_state, table],
+                    outputs=output_image)
+
+demo.launch()

requirements.txt

@@ -0,0 +1,7 @@
+gradio
+pillow
+pytesseract
+opencv-python
+matplotlib
+requests
+easyocr