split sample data for training and testing purposes

CNNmodel/image_proprocessing.py

+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+Author : Gunasinghe M.D.
+IT NUM : IT17043342
+
+"""
+
+# Image Preprocessing for train, test and validation sets
+
+import os
+import random
+import glob
+
+
+def prepare_test_data(n):
+    base_path = "shapes"
+    f1 = random.sample(glob.glob(os.path.join(base_path, "test/circles") + "/*"), n)
+    f2 = random.sample(glob.glob(os.path.join(base_path, "test/squares") + "/*"), n)
+    f3 = random.sample(glob.glob(os.path.join(base_path, "test/triangles") + "/*"), n)
+    for c in f1:
+        os.remove(c)
+    for s in f2:
+        os.remove(s)
+    for t in f3:
+        os.remove(t)
+
+def prepare_validation_data(n):
+    base_path = "shapes"
+    f1 = random.sample(glob.glob(os.path.join(base_path, "validation/circles") + "/*"), n)
+    f2 = random.sample(glob.glob(os.path.join(base_path, "validation/squares") + "/*"), n)
+    f3 = random.sample(glob.glob(os.path.join(base_path, "validation/triangles") + "/*"), n)
+    for c in f1:
+        os.remove(c)
+    for s in f2:
+        os.remove(s)
+    for t in f3:
+        os.remove(t)
+
+# n = number  of samples to remove from each categorical folder
+prepare_test_data(70)
+prepare_validation_data(40)
\ No newline at end of file