どこか間違えてるやつ

.gitignore

@@ -1,2 +1,2 @@
-"これ見えてない？"
 .idea
+.env

mepi_test.py

@@ -0,0 +1,49 @@
+import cv2
+import numpy as np
+
+# 画像読み込み
+image = cv2.imread("test2.png")
+image_gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
+
+# ノイズ軽減
+image_blur = cv2.GaussianBlur(image_gray, (5,5), 0)
+
+# Cannyエッジ
+edges = cv2.Canny(image_blur, 30, 100)
+
+# 線を太らせる
+kernel = np.ones((3,3), np.uint8)
+edges = cv2.dilate(edges, kernel, iterations=2)
+
+# 輪郭検出
+contours, _ = cv2.findContours(edges, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
+
+image_result = image.copy()
+
+# 四角形数をカウント
+square_count = 0
+
+# 白背景画像（描画用）
+image_blank = np.ones_like(image) * 255
+
+for cnt in contours:
+    area = cv2.contourArea(cnt)
+    if area < 5000 or area>20000:  # 小さい輪郭を除外
+        cv2.drawContours(image_result, [cnt], -1, (255,0,0), 2),
+        continue
+
+    peri = cv2.arcLength(cnt, True)
+    approx = cv2.approxPolyDP(cnt, 0.04 * peri, True)
+
+    if len(approx) == 4:
+        square_count += 1
+        cv2.drawContours(image_result, [approx], -1, (0,255,0), 2)
+    else:
+        cv2.drawContours(image_result, [approx], -1, (0,0,255), 2)
+
+
+# 結果保存
+cv2.imwrite('detected_squares.png', image_result)
+
+# 結果出力
+print(f"検出された四角形の数: {square_count}")

mepi_test_1.py

@@ -0,0 +1,48 @@
+from sklearn import svm
+import cv2
+import numpy as np
+import glob
+from joblib import dump
+
+# HOG
+hog = cv2.HOGDescriptor(
+    _winSize=(64,64),
+    _blockSize=(16,16),
+    _blockStride=(8,8),
+    _cellSize=(8,8),
+    _nbins=9
+)
+
+x = []
+y = []
+
+# データセット読み込み
+for label in ['fu_up','fu_down',
+              'tokinn_up','tokinn_down',
+              'ou_up','ou_down',
+              'gyoku_up','gyoku_down',
+              'keima_up','keima_down',
+              'None']:
+    for file in glob.glob(f'data/{label}/*.png'):
+        img = cv2.imread(file, cv2.IMREAD_GRAYSCALE)
+        img = cv2.resize(img, (64,64))
+        descriptor = hog.compute(img)
+        x.append(descriptor.flatten())
+        y.append(label)
+
+x = np.array(x)
+y = np.array(y)
+
+# SVM学習
+clf = svm.SVC(kernel='linear')
+clf.fit(X, y)
+dump(clf, 'shogi_model.joblib')
+
+
+img = cv2.imread('test_keima.png', cv2.IMREAD_GRAYSCALE)
+img = cv2.resize(img, (64,64))
+descriptor = hog.compute(img)
+predicted = clf.predict([descriptor.flatten()])
+
+print(f"推論結果: {predicted[0]}")
+