Create app.py

app.py

@@ -0,0 +1,128 @@
+import gradio as gr
+import cv2
+import numpy as np
+from PIL import Image, ImageDraw, ImageFont
+import tempfile
+import os
+
+def process_video_with_mask(video_path, mask_path, text="Joker"):
+    # 创建临时输出文件
+    output_path = os.path.join(tempfile.gettempdir(), "output.mp4")
+    
+    # 打开视频和mask视频
+    cap_video = cv2.VideoCapture(video_path)
+    cap_mask = cv2.VideoCapture(mask_path)
+    
+    # 获取视频参数
+    fps = cap_video.get(cv2.CAP_PROP_FPS)
+    width = int(cap_video.get(cv2.CAP_PROP_FRAME_WIDTH))
+    height = int(cap_video.get(cv2.CAP_PROP_FRAME_HEIGHT))
+    
+    # 创建视频写入器
+    fourcc = cv2.VideoWriter_fourcc(*'mp4v')
+    out = cv2.VideoWriter(output_path, fourcc, fps, (width, height))
+    
+    while True:
+        ret_video, frame_video = cap_video.read()
+        ret_mask, frame_mask = cap_mask.read()
+        
+        if not ret_video or not ret_mask:
+            break
+            
+        # 处理mask为单通道
+        if len(frame_mask.shape) == 3:
+            frame_mask = cv2.cvtColor(frame_mask, cv2.COLOR_BGR2GRAY)
+        
+        # 找到mask中的白色区域
+        _, binary = cv2.threshold(frame_mask, 200, 255, cv2.THRESH_BINARY)
+        
+        # 查找轮廓
+        contours, _ = cv2.findContours(binary, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+        
+        # 转换为PIL图像处理文字
+        pil_image = Image.fromarray(cv2.cvtColor(frame_video, cv2.COLOR_BGR2RGB))
+        draw = ImageDraw.Draw(pil_image)
+        
+        if contours:
+            # 找到面积最大的轮廓
+            max_contour = max(contours, key=cv2.contourArea)
+            
+            # 获取旋转矩形
+            rect = cv2.minAreaRect(max_contour)
+            box = cv2.boxPoints(rect)
+            box = np.int0(box)
+            
+            # 计算中心点和尺寸
+            center = rect[0]
+            size = rect[1]
+            angle = rect[2]
+            
+            # 创建透明画布绘制文字
+            text_image = Image.new("RGBA", (int(size[0]), int(size[1])), (0, 0, 0, 0))
+            text_draw = ImageDraw.Draw(text_image)
+            
+            try:
+                # 自动调整字体大小
+                font_size = int(min(size[0] * 0.8 / max(1, len(text)), size[1] * 0.8))
+                font = ImageFont.truetype("华文琥珀.ttf", font_size)
+            except:
+                font = ImageFont.load_default()
+            
+            # 计算文字位置
+            bbox = text_draw.textbbox((0, 0), text, font=font)
+            text_w = bbox[2] - bbox[0]
+            text_h = bbox[3] - bbox[1]
+            pos_x = (size[0] - text_w) // 2
+            pos_y = (size[1] - text_h) // 2
+            
+            # 绘制文字
+            text_draw.text((pos_x, pos_y), text, fill=(255, 255, 0, 255), font=font)
+            
+            # 旋转文字
+            rotated_text = text_image.rotate(angle, expand=True, center=(size[0]/2, size[1]/2))
+            
+            # 计算粘贴位置
+            paste_x = int(center[0] - rotated_text.width / 2)
+            paste_y = int(center[1] - rotated_text.height / 2)
+            
+            # 合并到原图
+            pil_image.paste(rotated_text, (paste_x, paste_y), rotated_text)
+        
+        # 转换回OpenCV格式
+        frame_processed = cv2.cvtColor(np.array(pil_image), cv2.COLOR_RGB2BGR)
+        out.write(frame_processed)
+    
+    # 释放资源
+    cap_video.release()
+    cap_mask.release()
+    out.release()
+    
+    return output_path
+
+# 创建Gradio界面
+with gr.Blocks() as demo:
+    gr.Markdown("## 视频文字替换工具")
+    with gr.Row():
+        with gr.Column():
+            video_input = gr.Video(label="原始视频")
+            mask_input = gr.Video(label="Mask视频")
+            text_input = gr.Textbox(label="要添加的文字", value="示例文字")
+            submit_btn = gr.Button("处理视频")
+        with gr.Column():
+            video_output = gr.Video(label="处理结果")
+    
+    submit_btn.click(
+        fn=process_video_with_mask,
+        inputs=[video_input, mask_input, text_input],
+        outputs=video_output,
+    )
+
+    gr.Examples(
+        [
+            ["examples/20250511-013442_3331156457_.mp4", "examples/20250511-013442_3331156457__temp.mp4_alpha_temp.mp4", "今天天气怎么样？"]
+        ],
+        [video_input, mask_input, text_input]
+    )
+
+if __name__ == "__main__":
+    demo.launch(share=True)