调了一个端木说gohome的还不错的参数

classifer.py

@@ -31,7 +31,7 @@ sys.path.append(os.getcwd())
 from root_dir import ROOT_DIR
 import utils
 
-FEATURE_INDEX = [0,1,2,4]
+FEATURE_INDEX = [0,1,2,3,4,5]
 delete_columns = 10  # 已弃用
 num_bins = 10
 
@@ -60,10 +60,10 @@ class WoodClass(object):
             self._single_pick = single_pick_mode
             self.set_purity(self.pur)
             self.change_pick_mode(single_pick_mode)
-            # self.model = LogisticRegression(C=1e5)
+            self.model = LogisticRegression(C=1e5)
             self.left_correct = left_correct
             # self.model = KNeighborsClassifier()
-            self.model = DecisionTreeClassifier()
+            # self.model = DecisionTreeClassifier()
         else:
             self.load(load_from)
         self.isCorrect = False
@@ -113,7 +113,7 @@ class WoodClass(object):
         :return:
         """
         # 训练数据文件位置
-        result = self.get_train_data(data_path, plot_2d=False)
+        result = self.get_train_data(data_path, plot_2d=True)
         if result is False:
             return 0
         x, y = result
@@ -122,7 +122,7 @@ class WoodClass(object):
         model_name = self.save(file_name)
         return model_name
 
-    def fit(self, x, y, test_size=0.7):
+    def fit(self, x, y, test_size=0.3):
         x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=test_size, random_state=0)
         self.model.fit(x_train, y_train)
         y_pred = self.model.predict(x_test)
@@ -304,14 +304,14 @@ class WoodClass(object):
         x = cv2.cvtColor(x, cv2.COLOR_BGR2LAB)
         x = np.concatenate((x, x_hsv), axis=2)
         x = np.reshape(x, (x.shape[0] * x.shape[1], x.shape[2]))
-
+        x = x[x[:, 0] > 30]
         # x = x[np.argsort(x[:, 0])]
         # x = x[-self.k:, :]
 
 
         hist, bins = np.histogram(x[:, 0], bins=num_bins)
-        hist = hist[1:]
-        bins = bins[1:]
+        # hist = hist[1:]
+        # bins = bins[1:]
         sorted_indices = np.argsort(hist)
         hist_number = sorted_indices[-1]
         second_hist_number = sorted_indices[-2]
@@ -440,8 +440,8 @@ class WoodClass(object):
         :param standard_color: 标准色彩, 默认为白色
         :return: 校正后的图片 shape = (n_rows, n_cols - cut_col_num, n_channels)
         """
-        if self.left_correct:
         img = img[:, 20:, :]  # 图片黑边需要去除
+        if self.left_correct:
             # 按照correct_col_num列数量取出最左侧校正板区域成像结果
             correct_img = img[:, :correct_col_num, :]
             # 校正区域进行均值化
@@ -468,7 +468,7 @@ if __name__ == '__main__':
     settings.model_path = str(ROOT_DIR / 'models' / wood.fit_pictures(data_path=data_path))
 
     # 测试单张图片的预测，predict_mode=True表示导入本地的model, False为现场训练的
-    pic = cv2.imread(r"data/316/dark/rgb70.png")
+    pic = cv2.imread(r"data/318/dark/rgb89.png")
     start_time = time.time()
     # for i in range(100):
     wood_color = wood.predict(pic)

hist.py

@@ -13,7 +13,7 @@ ratio = np.sqrt(5000 / (w * h))
 ww, hh = int(ratio * w), int(ratio * h)
 img = cv2.resize(img, (hh, ww))
 x = img.reshape(img.shape[0]*img.shape[1], img.shape[2])
-
+x = x[x[:,0]>20]
 hist, bins = np.histogram(x[:, 0], bins=10)
 hist = hist[1:]
 bins = bins[1:]