batch images

.idea/workspace.xml

@@ -4,7 +4,12 @@
     <option name="autoReloadType" value="SELECTIVE" />
   </component>
   <component name="ChangeListManager">
-    <list default="true" id="5e4481c0-7ba2-42e4-bbe6-4c36a0d36baa" name="Changes" comment="remove dotenv" />
+    <list default="true" id="5e4481c0-7ba2-42e4-bbe6-4c36a0d36baa" name="Changes" comment="font">
+      <change beforePath="$PROJECT_DIR$/.idea/workspace.xml" beforeDir="false" afterPath="$PROJECT_DIR$/.idea/workspace.xml" afterDir="false" />
+      <change beforePath="$PROJECT_DIR$/CDM/detect_classify/classification.py" beforeDir="false" afterPath="$PROJECT_DIR$/CDM/detect_classify/classification.py" afterDir="false" />
+      <change beforePath="$PROJECT_DIR$/app.py" beforeDir="false" afterPath="$PROJECT_DIR$/app.py" afterDir="false" />
+      <change beforePath="$PROJECT_DIR$/logs/app.log" beforeDir="false" afterPath="$PROJECT_DIR$/logs/app.log" afterDir="false" />
+    </list>
     <option name="SHOW_DIALOG" value="false" />
     <option name="HIGHLIGHT_CONFLICTS" value="true" />
     <option name="HIGHLIGHT_NON_ACTIVE_CHANGELIST" value="false" />
@@ -99,14 +104,7 @@
       <workItem from="1723386970626" duration="451000" />
       <workItem from="1723387453009" duration="27328000" />
       <workItem from="1723978405562" duration="51637000" />
-      <workItem from="1725925225487" duration="135179000" />
-    </task>
-    <task id="LOCAL-00011" summary="google drive">
-      <created>1723456219009</created>
-      <option name="number" value="00011" />
-      <option name="presentableId" value="LOCAL-00011" />
-      <option name="project" value="LOCAL" />
-      <updated>1723456219009</updated>
+      <workItem from="1725925225487" duration="136670000" />
     </task>
     <task id="LOCAL-00012" summary="google drive">
       <created>1723456794848</created>
@@ -444,7 +442,14 @@
       <option name="project" value="LOCAL" />
       <updated>1728915832031</updated>
     </task>
-    <option name="localTasksCounter" value="60" />
+    <task id="LOCAL-00060" summary="font">
+      <created>1728916518798</created>
+      <option name="number" value="00060" />
+      <option name="presentableId" value="LOCAL-00060" />
+      <option name="project" value="LOCAL" />
+      <updated>1728916518798</updated>
+    </task>
+    <option name="localTasksCounter" value="61" />
     <servers />
   </component>
   <component name="TypeScriptGeneratedFilesManager">
@@ -481,9 +486,10 @@
     <MESSAGE value="output img type" />
     <MESSAGE value="." />
     <MESSAGE value="remove dotenv" />
-    <option name="LAST_COMMIT_MESSAGE" value="remove dotenv" />
+    <MESSAGE value="font" />
+    <option name="LAST_COMMIT_MESSAGE" value="font" />
   </component>
   <component name="com.intellij.coverage.CoverageDataManagerImpl">
-    <SUITE FILE_PATH="coverage/Cpp4App_test$app.coverage" NAME="app Coverage Results" MODIFIED="1728910172560" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="true" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$" />
+    <SUITE FILE_PATH="coverage/Cpp4App_test$app.coverage" NAME="app Coverage Results" MODIFIED="1728918751288" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="true" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$" />
   </component>
 </project>

CDM/detect_classify/classification.py

@@ -209,10 +209,6 @@ def compo_classification(input_img, output_root, segment_root, merge_json, outpu
 
     # --------- classification ----------
 
-    classification_start_time = time.process_time()
-
-    model = get_clf_model(clf_model)
-
     # for compo in compos:
     #
     #     # comp_grey = grey[compo.row_min:compo.row_max, compo.col_min:compo.col_max]
@@ -290,70 +286,131 @@ def compo_classification(input_img, output_root, segment_root, merge_json, outpu
     #     else:
     #         print("clf_model has to be ResNet18 or ViT")
 
-    for compo in compos:
-        compo_start_time = time.process_time()
+# =============================================================================
+
+    # classification_start_time = time.process_time()
+    #
+    # model = get_clf_model(clf_model)
+    #
+    # for compo in compos:
+    #     compo_start_time = time.process_time()
+    #
+    #     # 计时预处理部分
+    #     preprocess_start = time.process_time()
+    #     comp_grey = grey[compo.row_min:compo.row_max, compo.col_min:compo.col_max]
+    #     preprocess_time = time.process_time() - preprocess_start
+    #
+    #     # 计时图像调整部分
+    #     resize_start = time.process_time()
+    #     if clf_model == "ResNet18":
+    #         comp_crop = cv2.resize(comp_grey, (32, 32))
+    #     elif clf_model == "ViT":
+    #         comp_crop = cv2.resize(comp_grey, (224, 224))
+    #     resize_time = time.process_time() - resize_start
+    #
+    #     # 计时张量转换部分
+    #     tensor_start = time.process_time()
+    #     if clf_model == "ResNet18":
+    #         comp_crop = comp_crop.reshape(1, 1, 32, 32)
+    #         comp_tensor = torch.tensor(comp_crop)
+    #         comp_tensor = comp_tensor.permute(0, 1, 3, 2)
+    #     elif clf_model == "ViT":
+    #         comp_tensor = torch.from_numpy(comp_crop)
+    #         comp_tensor = comp_tensor.view(1, 224, 224).repeat(3, 1, 1)
+    #         comp_tensor = comp_tensor.unsqueeze(0)
+    #     tensor_time = time.process_time() - tensor_start
+    #
+    #     # 计时模型推理部分
+    #     inference_start = time.process_time()
+    #     with torch.no_grad():
+    #         if clf_model == "ResNet18":
+    #             pred_label = model(comp_tensor)
+    #             output = pred_label
+    #         elif clf_model == "ViT":
+    #             output = model(comp_tensor)
+    #     inference_time = time.process_time() - inference_start
+    #
+    #     # 计时后处理部分
+    #     postprocess_start = time.process_time()
+    #     if clf_model == "ResNet18":
+    #         predicted = np.argmax(output.cpu().data.numpy(), axis=1)[0]
+    #     elif clf_model == "ViT":
+    #         _, predicted = torch.max(output.logits, 1)
+    #         predicted = predicted.cpu().numpy()[0]
+    #
+    #     if str(predicted) in label_dic.keys():
+    #         compo.label = label_dic[str(predicted)]
+    #         elements.append(compo)
+    #     else:
+    #         compo.label = str(predicted)
+    #     postprocess_time = time.process_time() - postprocess_start
+    #
+    #     compo_total_time = time.process_time() - compo_start_time
+    #
+    #     # 输出每个部分的耗时
+    #     print("==============================================")
+    #     print(f"Component processing time: {compo_total_time:.4f}s")
+    #     print(f"  Preprocessing time: {preprocess_time:.4f}s")
+    #     print(f"  Resize time: {resize_time:.4f}s")
+    #     print(f"  Tensor conversion time: {tensor_time:.4f}s")
+    #     print(f"  Inference time: {inference_time:.4f}s")
+    #     print(f"  Post-processing time: {postprocess_time:.4f}s\n")
+    #     print("==============================================")
+
+# =============================================================================
 
-        # 计时预处理部分
-        preprocess_start = time.process_time()
+    classification_start_time = time.process_time()
+
+    model = get_clf_model(clf_model)
+    comp_tensors = []
+    elements = []
+
+    # 收集所有张量
+    for compo in compos:
+        # 预处理
         comp_grey = grey[compo.row_min:compo.row_max, compo.col_min:compo.col_max]
-        preprocess_time = time.process_time() - preprocess_start
 
-        # 计时图像调整部分
-        resize_start = time.process_time()
+        # 调整图像大小
         if clf_model == "ResNet18":
             comp_crop = cv2.resize(comp_grey, (32, 32))
         elif clf_model == "ViT":
             comp_crop = cv2.resize(comp_grey, (224, 224))
-        resize_time = time.process_time() - resize_start
 
-        # 计时张量转换部分
-        tensor_start = time.process_time()
+        # 张量转换
         if clf_model == "ResNet18":
             comp_crop = comp_crop.reshape(1, 1, 32, 32)
-            comp_tensor = torch.tensor(comp_crop)
-            comp_tensor = comp_tensor.permute(0, 1, 3, 2)
+            comp_tensor = torch.tensor(comp_crop).permute(0, 1, 3, 2)
         elif clf_model == "ViT":
             comp_tensor = torch.from_numpy(comp_crop)
-            comp_tensor = comp_tensor.view(1, 224, 224).repeat(3, 1, 1)
-            comp_tensor = comp_tensor.unsqueeze(0)
-        tensor_time = time.process_time() - tensor_start
-
-        # 计时模型推理部分
-        inference_start = time.process_time()
-        with torch.no_grad():
-            if clf_model == "ResNet18":
-                pred_label = model(comp_tensor)
-                output = pred_label
-            elif clf_model == "ViT":
-                output = model(comp_tensor)
-        inference_time = time.process_time() - inference_start
-
-        # 计时后处理部分
-        postprocess_start = time.process_time()
+            comp_tensor = comp_tensor.view(1, 224, 224).repeat(3, 1, 1).unsqueeze(0)
+
+        comp_tensors.append(comp_tensor)
+
+    # 将张量堆叠成批次
+    batch_tensor = torch.cat(comp_tensors, dim=0)
+
+    # 模型推理
+    with torch.no_grad():
         if clf_model == "ResNet18":
-            predicted = np.argmax(output.cpu().data.numpy(), axis=1)[0]
+            output = model(batch_tensor)
         elif clf_model == "ViT":
-            _, predicted = torch.max(output.logits, 1)
-            predicted = predicted.cpu().numpy()[0]
+            output = model(batch_tensor)
 
-        if str(predicted) in label_dic.keys():
-            compo.label = label_dic[str(predicted)]
+    # 后处理
+    if clf_model == "ResNet18":
+        predicted = np.argmax(output.cpu().numpy(), axis=1)
+    elif clf_model == "ViT":
+        _, predicted = torch.max(output.logits, 1)
+        predicted = predicted.cpu().numpy()
+
+    # 为组件分配标签
+    for idx, compo in enumerate(compos):
+        pred_label = predicted[idx]
+        if str(pred_label) in label_dic.keys():
+            compo.label = label_dic[str(pred_label)]
             elements.append(compo)
         else:
-            compo.label = str(predicted)
-        postprocess_time = time.process_time() - postprocess_start
-
-        compo_total_time = time.process_time() - compo_start_time
-
-        # 输出每个部分的耗时
-        print("==============================================")
-        print(f"Component processing time: {compo_total_time:.4f}s")
-        print(f"  Preprocessing time: {preprocess_time:.4f}s")
-        print(f"  Resize time: {resize_time:.4f}s")
-        print(f"  Tensor conversion time: {tensor_time:.4f}s")
-        print(f"  Inference time: {inference_time:.4f}s")
-        print(f"  Post-processing time: {postprocess_time:.4f}s\n")
-        print("==============================================")
+            compo.label = str(pred_label)
 
     time_cost_ic = time.process_time() - classification_start_time
     print("time cost for icon classification: %2.2f s" % time_cost_ic)

app.py

@@ -29,7 +29,7 @@ from googleapiclient.http import MediaFileUpload
 # from dotenv import load_dotenv
 import os
 
-# 加载 .env 文件中的环境变量
+# # 加载 .env 文件中的环境变量
 # load_dotenv()
 
 title = "Cpp4App_test"

logs/app.log

@@ -0,0 +1,7 @@
+2024-10-15 02:12:39 - INFO - Application started
+2024-10-15 02:12:39 - INFO - file_cache is only supported with oauth2client<4.0.0
+2024-10-15 02:12:39 - INFO - HTTP Request: GET http://127.0.0.1:7860/startup-events "HTTP/1.1 200 OK"
+2024-10-15 02:12:39 - INFO - HTTP Request: HEAD http://127.0.0.1:7860/ "HTTP/1.1 200 OK"
+2024-10-15 02:12:40 - INFO - HTTP Request: GET https://checkip.amazonaws.com/ "HTTP/1.1 200 "
+2024-10-15 02:12:40 - INFO - HTTP Request: GET https://api.gradio.app/pkg-version "HTTP/1.1 200 OK"
+2024-10-15 02:12:40 - INFO - HTTP Request: GET https://checkip.amazonaws.com/ "HTTP/1.1 200 "