upload all examples

.gitattributes

@@ -34,3 +34,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 *.png filter=lfs diff=lfs merge=lfs -text
+*.jpg filter=lfs diff=lfs merge=lfs -text

app.py

@@ -7,16 +7,16 @@ import random
 # 图像增强
 ENHANCE_INPUT_DIR = "sr_input"
 ENHANCE_OUTPUT_DIR = "sr_output"
-ENHANCE_IMG_NAMES = [f"{i}.png" for i in range(1, 30)]
+ENHANCE_IMG_NAMES = [f"img_{i}.jpg" for i in range(1, 30)]
 
 # 分割
 SEG_INPUT_DIR = "seg_input"
 SEG_OUTPUT_DIR = "seg_output"
-SEG_IMG_NAMES = [f"{i}.png" for i in range(1, 30)]
+SEG_IMG_NAMES = [f"img_{i}.png" for i in range(1, 30)]
 
 # 诊断
 DIAG_INPUT_DIR = "diag_input"
-DIAG_IMG_NAMES = [f"{i}.png" for i in range(1, 30)]
+DIAG_IMG_NAMES = [f"{i}.png" for i in range(1, 32)]
 
 # Kvasir 类别
 KVASIR_CLASSES = [
@@ -28,7 +28,9 @@ KVASIR_CLASSES = [
     "糜烂 (dyed-lifted-polyps)",
     "出血 (dyed-resection-margins)",
     "炎症 (bbps-0-1)",
-    "直肠癌 (bbps-2-3)"
+    "直肠癌 (bbps-2-3)",
+    "normal-cecum",
+    "normal-pylorus"
 ]
 
 # ========== 功能函数 ==========
@@ -59,9 +61,46 @@ def show_diag_input(img_name):
     return open_img(path)
 
 def diagnose_demo(img_name):
-    pred = random.choice(KVASIR_CLASSES)
-    conf = round(random.uniform(0.85, 0.99), 2)
-    return f"诊断结果：{pred}\n置信度：{conf}"
+    idx = int(os.path.splitext(img_name)[0])
+    # 区间与类别映射（1-based index）
+    idx2cat = [
+        (range(1, 5), "糜烂 (dyed-lifted-polyps)"),
+        (range(5, 9), "出血 (dyed-resection-margins)"),
+        (range(9, 13), "食管静脉曲张 (esophagitis)"),
+        (range(13, 17), "normal-cecum"),
+        (range(17, 21), "normal-pylorus"),
+        (range(21, 25), "正常粘膜 (normal-z-line)"),
+        (range(25, 29), "息肉 (polyps)"),
+        (range(29, 33), "溃疡 (ulcerative-colitis)")
+    ]
+    main_cat = None
+    for r, cat in idx2cat:
+        if idx in r:
+            main_cat = cat
+            break
+    # 若没落在区间，随机主类别
+    if main_cat is None:
+        main_cat = random.choice(KVASIR_CLASSES)
+    # 概率分配
+    n_cat = len(KVASIR_CLASSES)
+    probs = [0.0 for _ in range(n_cat)]
+    main_idx = KVASIR_CLASSES.index(main_cat)
+    main_prob = random.uniform(0.6, 0.8)
+    rest = 1 - main_prob
+    # 其余类别分配小随机
+    other_probs_raw = [random.uniform(0.01, 1) for _ in range(n_cat-1)]
+    other_probs_norm = [x / sum(other_probs_raw) * rest for x in other_probs_raw]
+    pidx = 0
+    for i in range(n_cat):
+        if i == main_idx:
+            probs[i] = main_prob
+        else:
+            probs[i] = other_probs_norm[pidx]
+            pidx += 1
+    # 最终输出
+    result_table = [[KVASIR_CLASSES[i], f"{probs[i]:.4f}"] for i in range(n_cat)]
+    result_text = f"诊断类别：{main_cat}（概率最大）"
+    return result_table, result_text
 
 # ========== Gradio 布局 ==========
 with gr.Blocks(title="AI内窥镜影像辅助分析系统") as demo:
@@ -77,9 +116,7 @@ with gr.Blocks(title="AI内窥镜影像辅助分析系统") as demo:
                 enhance_output_img = gr.Image(label="增强结果图片", interactive=False)
             enhance_btn = gr.Button("增强图片")
 
-            # 选择图片自动显示原图
             enhance_select.change(fn=show_enhance_input, inputs=enhance_select, outputs=enhance_input_img)
-            # 按钮显示增强结果
             enhance_btn.click(fn=enhance_demo, inputs=enhance_select, outputs=enhance_output_img)
 
     # 下方：左右两大模块
@@ -101,9 +138,10 @@ with gr.Blocks(title="AI内窥镜影像辅助分析系统") as demo:
             diag_select = gr.Dropdown(choices=DIAG_IMG_NAMES, value=DIAG_IMG_NAMES[0], label="选择样例图片", filterable=True)
             diag_input_img = gr.Image(label="输入图片", interactive=False)
             diag_btn = gr.Button("开始诊断")
+            diag_table = gr.Dataframe(headers=["类别", "概率"])
             diag_result = gr.Textbox(label="诊断结果")
 
             diag_select.change(fn=show_diag_input, inputs=diag_select, outputs=diag_input_img)
-            diag_btn.click(fn=diagnose_demo, inputs=diag_select, outputs=diag_result)
+            diag_btn.click(fn=diagnose_demo, inputs=diag_select, outputs=[diag_table, diag_result])
 
 demo.launch()