100 triangles, 100 squares, and 100 circles in processing. each png image is 28×28 px, the images are in 3 folders labeled squares, circles, and triangles pretty straightforward
We have to find the shape that falls in its category by training the model using Neural Networks.
Importing Libraries
Unzipping the data
Data Preprocessing
Checking the number of content inside the data by making result, image, and files variable.
300
300
[\'shapes/triangles/drawing(3).png\', \'shapes/triangles/drawing(16).png\', \'shapes/triangles/drawing(71).png\', \'shapes/triangles/drawing(68).png\', \'shapes/triangles/drawing(75).png\']
300
Data Visualization
Visually checking the data in the form of tables
Shuffling the dataset
Assigning values to category of shapes
Dropping Useless Column
Type Conversion
Train Test Split
Splitting the data in 80–20 split
Model Architecture
Model: "sequential" _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= conv2d (Conv2D) (None, 28, 28, 32) 320 _________________________________________________________________ conv2d_1 (Conv2D) (None, 26, 26, 32) 9248 _________________________________________________________________ max_pooling2d (MaxPooling2D) (None, 13, 13, 32) 0 _________________________________________________________________ dropout (Dropout) (None, 13, 13, 32) 0 _________________________________________________________________ conv2d_2 (Conv2D) (None, 11, 11, 64) 18496 _________________________________________________________________ max_pooling2d_1 (MaxPooling2 (None, 5, 5, 64) 0 _________________________________________________________________ flatten (Flatten) (None, 1600) 0 _________________________________________________________________ dense (Dense) (None, 64) 102464 _________________________________________________________________ dropout_1 (Dropout) (None, 64) 0 _________________________________________________________________ dense_1 (Dense) (None, 3) 195 ================================================================= Total params: 130,723 Trainable params: 130,723 Non-trainable params: 0 _________________________________________________________________
Training the model
Epoch 1/100 12/12 [==============================] - 0s 17ms/step - loss: 1.1123 - accuracy: 0.3042 - val_loss: 1.0959 - val_accuracy: 0.3167 Epoch 2/100 12/12 [==============================] - 0s 5ms/step - loss: 1.0974 - accuracy: 0.3417 - val_loss: 1.0921 - val_accuracy: 0.4333 Epoch 3/100 12/12 [==============================] - 0s 4ms/step - loss: 1.0963 - accuracy: 0.3333 - val_loss: 1.0907 - val_accuracy: 0.3500 Epoch 4/100 12/12 [==============================] - 0s 5ms/step - loss: 1.0819 - accuracy: 0.4333 - val_loss: 1.0654 - val_accuracy: 0.5167 Epoch 5/100 12/12 [==============================] - 0s 4ms/step - loss: 1.0621 - accuracy: 0.4542 - val_loss: 1.0147 - val_accuracy: 0.4833 Epoch 6/100 12/12 [==============================] - 0s 6ms/step - loss: 0.9976 - accuracy: 0.5167 - val_loss: 0.9249 - val_accuracy: 0.5333 Epoch 7/100 12/12 [==============================] - 0s 6ms/step - loss: 0.9264 - accuracy: 0.5917 - val_loss: 0.8490 - val_accuracy: 0.5833 Epoch 8/100 12/12 [==============================] - 0s 5ms/step - loss: 0.8427 - accuracy: 0.5750 - val_loss: 0.7379 - val_accuracy: 0.7000 Epoch 9/100 12/12 [==============================] - 0s 5ms/step - loss: 0.7102 - accuracy: 0.7292 - val_loss: 0.6010 - val_accuracy: 0.8167 Epoch 10/100 12/12 [==============================] - 0s 4ms/step - loss: 0.5572 - accuracy: 0.7917 - val_loss: 0.4865 - val_accuracy: 0.8167 Epoch 11/100 12/12 [==============================] - 0s 5ms/step - loss: 0.5107 - accuracy: 0.8167 - val_loss: 0.4676 - val_accuracy: 0.8500 Epoch 12/100 12/12 [==============================] - 0s 4ms/step - loss: 0.3793 - accuracy: 0.8833 - val_loss: 0.4695 - val_accuracy: 0.7833 Epoch 13/100 12/12 [==============================] - 0s 4ms/step - loss: 0.3837 - accuracy: 0.8583 - val_loss: 0.3716 - val_accuracy: 0.8333 Epoch 14/100 12/12 [==============================] - 0s 4ms/step - loss: 0.3282 - accuracy: 0.8833 - val_loss: 0.3677 - val_accuracy: 0.8500 Epoch 15/100 12/12 [==============================] - 0s 5ms/step - loss: 0.2777 - accuracy: 0.9125 - val_loss: 0.3662 - val_accuracy: 0.8667 Epoch 16/100 12/12 [==============================] - 0s 4ms/step - loss: 0.2536 - accuracy: 0.9208 - val_loss: 0.3261 - val_accuracy: 0.8667 Epoch 17/100 12/12 [==============================] - 0s 3ms/step - loss: 0.2008 - accuracy: 0.9458 - val_loss: 0.3115 - val_accuracy: 0.8500 Epoch 18/100 12/12 [==============================] - 0s 6ms/step - loss: 0.1649 - accuracy: 0.9458 - val_loss: 0.3499 - val_accuracy: 0.8667 Epoch 19/100 12/12 [==============================] - 0s 5ms/step - loss: 0.1265 - accuracy: 0.9583 - val_loss: 0.3479 - val_accuracy: 0.8833 Epoch 20/100 12/12 [==============================] - 0s 4ms/step - loss: 0.1588 - accuracy: 0.9458 - val_loss: 0.3476 - val_accuracy: 0.8667 Epoch 21/100 12/12 [==============================] - 0s 3ms/step - loss: 0.1662 - accuracy: 0.9458 - val_loss: 0.3211 - val_accuracy: 0.8667 Epoch 22/100 12/12 [==============================] - 0s 3ms/step - loss: 0.1309 - accuracy: 0.9417 - val_loss: 0.3586 - val_accuracy: 0.8500 Epoch 23/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0895 - accuracy: 0.9625 - val_loss: 0.3904 - val_accuracy: 0.8500 Epoch 24/100 12/12 [==============================] - 0s 3ms/step - loss: 0.1067 - accuracy: 0.9625 - val_loss: 0.5581 - val_accuracy: 0.8167 Epoch 25/100 12/12 [==============================] - 0s 6ms/step - loss: 0.1033 - accuracy: 0.9667 - val_loss: 0.3578 - val_accuracy: 0.9000 Epoch 26/100 12/12 [==============================] - 0s 4ms/step - loss: 0.0626 - accuracy: 0.9917 - val_loss: 0.3928 - val_accuracy: 0.8833 Epoch 27/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0536 - accuracy: 0.9917 - val_loss: 0.3721 - val_accuracy: 0.9000 Epoch 28/100 12/12 [==============================] - 0s 4ms/step - loss: 0.0527 - accuracy: 0.9875 - val_loss: 0.4973 - val_accuracy: 0.8667 Epoch 29/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0808 - accuracy: 0.9708 - val_loss: 0.4575 - val_accuracy: 0.8833 Epoch 30/100 12/12 [==============================] - 0s 3ms/step - loss: 0.1053 - accuracy: 0.9667 - val_loss: 0.4273 - val_accuracy: 0.8667 Epoch 31/100 12/12 [==============================] - 0s 5ms/step - loss: 0.0559 - accuracy: 0.9958 - val_loss: 0.4082 - val_accuracy: 0.9167 Epoch 32/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0435 - accuracy: 0.9875 - val_loss: 0.5094 - val_accuracy: 0.8667 Epoch 33/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0429 - accuracy: 0.9917 - val_loss: 0.4551 - val_accuracy: 0.9167 Epoch 34/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0357 - accuracy: 0.9917 - val_loss: 0.5260 - val_accuracy: 0.8833 Epoch 35/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0365 - accuracy: 0.9958 - val_loss: 0.4955 - val_accuracy: 0.9000 Epoch 36/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0282 - accuracy: 1.0000 - val_loss: 0.3949 - val_accuracy: 0.9167 Epoch 37/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0303 - accuracy: 0.9875 - val_loss: 0.4732 - val_accuracy: 0.9167 Epoch 38/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0169 - accuracy: 1.0000 - val_loss: 0.4634 - val_accuracy: 0.9000 Epoch 39/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0188 - accuracy: 0.9958 - val_loss: 0.4746 - val_accuracy: 0.9167 Epoch 40/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0140 - accuracy: 0.9958 - val_loss: 0.5232 - val_accuracy: 0.8833 Epoch 41/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0119 - accuracy: 0.9958 - val_loss: 0.4813 - val_accuracy: 0.9167 Epoch 42/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0168 - accuracy: 0.9958 - val_loss: 0.4460 - val_accuracy: 0.9167 Epoch 43/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0163 - accuracy: 0.9958 - val_loss: 0.4505 - val_accuracy: 0.9167 Epoch 44/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0164 - accuracy: 0.9958 - val_loss: 0.4882 - val_accuracy: 0.9167 Epoch 45/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0250 - accuracy: 0.9917 - val_loss: 0.5581 - val_accuracy: 0.9167 Epoch 46/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0245 - accuracy: 0.9958 - val_loss: 0.4527 - val_accuracy: 0.9167 Epoch 47/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0253 - accuracy: 0.9917 - val_loss: 0.4838 - val_accuracy: 0.9000 Epoch 48/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0148 - accuracy: 0.9958 - val_loss: 0.4740 - val_accuracy: 0.9000 Epoch 49/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0104 - accuracy: 1.0000 - val_loss: 0.4538 - val_accuracy: 0.9167 Epoch 50/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0127 - accuracy: 0.9958 - val_loss: 0.5382 - val_accuracy: 0.9000 Epoch 51/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0139 - accuracy: 0.9958 - val_loss: 0.4694 - val_accuracy: 0.9167 Epoch 52/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0361 - accuracy: 0.9917 - val_loss: 0.4307 - val_accuracy: 0.9167 Epoch 53/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0125 - accuracy: 0.9958 - val_loss: 0.5136 - val_accuracy: 0.9167 Epoch 54/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0076 - accuracy: 1.0000 - val_loss: 0.4797 - val_accuracy: 0.9000 Epoch 55/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0231 - accuracy: 0.9875 - val_loss: 0.5369 - val_accuracy: 0.9167 Epoch 56/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0115 - accuracy: 1.0000 - val_loss: 0.5223 - val_accuracy: 0.9167 Epoch 57/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0115 - accuracy: 0.9958 - val_loss: 0.6016 - val_accuracy: 0.9000 Epoch 58/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0081 - accuracy: 1.0000 - val_loss: 0.7670 - val_accuracy: 0.8500 Epoch 59/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0194 - accuracy: 0.9958 - val_loss: 0.5915 - val_accuracy: 0.9167 Epoch 60/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0074 - accuracy: 1.0000 - val_loss: 0.5328 - val_accuracy: 0.9000 Epoch 61/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0078 - accuracy: 1.0000 - val_loss: 0.5192 - val_accuracy: 0.9167 Epoch 62/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0102 - accuracy: 0.9958 - val_loss: 0.5077 - val_accuracy: 0.9167 Epoch 63/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0063 - accuracy: 1.0000 - val_loss: 0.5320 - val_accuracy: 0.9000 Epoch 64/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0095 - accuracy: 0.9958 - val_loss: 0.6002 - val_accuracy: 0.8833 Epoch 65/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0232 - accuracy: 0.9958 - val_loss: 1.0127 - val_accuracy: 0.8333 Epoch 66/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0255 - accuracy: 0.9917 - val_loss: 0.5754 - val_accuracy: 0.9167 Epoch 67/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0201 - accuracy: 0.9917 - val_loss: 0.5047 - val_accuracy: 0.8833 Epoch 68/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0189 - accuracy: 0.9958 - val_loss: 0.6710 - val_accuracy: 0.8833 Epoch 69/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0140 - accuracy: 1.0000 - val_loss: 0.5965 - val_accuracy: 0.9000 Epoch 70/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0070 - accuracy: 1.0000 - val_loss: 0.5595 - val_accuracy: 0.9167 Epoch 71/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0087 - accuracy: 1.0000 - val_loss: 0.5850 - val_accuracy: 0.8833 Epoch 72/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0078 - accuracy: 1.0000 - val_loss: 0.5876 - val_accuracy: 0.9167 Epoch 73/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0120 - accuracy: 0.9958 - val_loss: 0.6837 - val_accuracy: 0.9167 Epoch 74/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0056 - accuracy: 1.0000 - val_loss: 0.6401 - val_accuracy: 0.9000 Epoch 75/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0105 - accuracy: 0.9958 - val_loss: 0.6296 - val_accuracy: 0.9000 Epoch 76/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0069 - accuracy: 1.0000 - val_loss: 0.5819 - val_accuracy: 0.9000 Epoch 77/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0051 - accuracy: 1.0000 - val_loss: 0.5993 - val_accuracy: 0.9167 Epoch 78/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0017 - accuracy: 1.0000 - val_loss: 0.6399 - val_accuracy: 0.9167 Epoch 79/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0117 - accuracy: 0.9958 - val_loss: 0.7028 - val_accuracy: 0.9000 Epoch 80/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0067 - accuracy: 1.0000 - val_loss: 0.6796 - val_accuracy: 0.8833 Epoch 81/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0148 - accuracy: 0.9958 - val_loss: 0.8375 - val_accuracy: 0.8667 Epoch 82/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0425 - accuracy: 0.9833 - val_loss: 1.1823 - val_accuracy: 0.8167 Epoch 83/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0220 - accuracy: 0.9958 - val_loss: 0.7019 - val_accuracy: 0.9000 Epoch 84/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0039 - accuracy: 1.0000 - val_loss: 0.8417 - val_accuracy: 0.8667 Epoch 85/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0077 - accuracy: 1.0000 - val_loss: 0.7788 - val_accuracy: 0.8500 Epoch 86/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0054 - accuracy: 1.0000 - val_loss: 0.7818 - val_accuracy: 0.9000 Epoch 87/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0123 - accuracy: 0.9958 - val_loss: 0.7559 - val_accuracy: 0.9000 Epoch 88/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0090 - accuracy: 0.9958 - val_loss: 0.7779 - val_accuracy: 0.9000 Epoch 89/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0159 - accuracy: 0.9958 - val_loss: 0.9408 - val_accuracy: 0.8500 Epoch 90/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0065 - accuracy: 1.0000 - val_loss: 0.7213 - val_accuracy: 0.9167 Epoch 91/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0096 - accuracy: 1.0000 - val_loss: 0.6934 - val_accuracy: 0.9000 Epoch 92/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0045 - accuracy: 1.0000 - val_loss: 0.7442 - val_accuracy: 0.9167 Epoch 93/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0154 - accuracy: 0.9958 - val_loss: 0.7858 - val_accuracy: 0.9167 Epoch 94/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0046 - accuracy: 1.0000 - val_loss: 0.8481 - val_accuracy: 0.9000 Epoch 95/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0040 - accuracy: 1.0000 - val_loss: 0.7743 - val_accuracy: 0.9167 Epoch 96/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0138 - accuracy: 0.9958 - val_loss: 0.7438 - val_accuracy: 0.9000 Epoch 97/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0088 - accuracy: 1.0000 - val_loss: 0.7356 - val_accuracy: 0.8833 Epoch 98/100 12/12 [==============================] - 0s 5ms/step - loss: 0.0086 - accuracy: 0.9958 - val_loss: 0.8468 - val_accuracy: 0.9333 Epoch 99/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0112 - accuracy: 0.9917 - val_loss: 0.8711 - val_accuracy: 0.9333 Epoch 100/100 12/12 [==============================] - 0s 3ms/step - loss: 0.0091 - accuracy: 0.9958 - val_loss: 0.8142 - val_accuracy: 0.9167
Result and Prediction
[0.23137156665325165, 0.9833333492279053]
array([1, 0, 1, 1, 0, 0, 1, 2, 2, 2, 1, 0, 2, 1, 2, 1, 2, 2, 0, 2, 1, 0,
1, 0, 0, 2, 1, 1, 2, 1, 1, 2, 2, 0, 0, 0, 1, 0, 1, 0, 0, 2, 1, 1,
1, 1, 0, 1, 2, 0, 2, 2, 2, 1, 2, 1, 0, 2, 0, 2])
Accuracy and Loss Graph
Testing
array([0., 1., 0.], dtype=float32)
1
DeepCC
[INFO] Reading [keras model] \'best.h5\' [SUCCESS] Saved \'best_deepC/best.onnx\' [INFO] Reading [onnx model] \'best_deepC/best.onnx\' [INFO] Model info: ir_vesion : 5 doc : [WARNING] [ONNX]: graph-node conv2d\'s attribute auto_pad has no meaningful data. [WARNING] [ONNX]: terminal (input/output) conv2d_input\'s shape is less than 1. Changing it to 1. [WARNING] [ONNX]: terminal (input/output) dense_1\'s shape is less than 1. Changing it to 1. WARN (GRAPH): found operator node with the same name (dense_1) as io node. [INFO] Running DNNC graph sanity check ... [SUCCESS] Passed sanity check. [INFO] Writing C++ file \'best_deepC/best.cpp\' [INFO] deepSea model files are ready in \'best_deepC/\' [RUNNING COMMAND] g++ -std=c++11 -O3 -fno-rtti -fno-exceptions -I. -I/opt/tljh/user/lib/python3.7/site-packages/deepC-0.13-py3.7-linux-x86_64.egg/deepC/include -isystem /opt/tljh/user/lib/python3.7/site-packages/deepC-0.13-py3.7-linux-x86_64.egg/deepC/packages/eigen-eigen-323c052e1731 "best_deepC/best.cpp" -D_AITS_MAIN -o "best_deepC/best.exe" [RUNNING COMMAND] size "best_deepC/best.exe" text data bss dec hex filename 690061 3784 760 694605 a994d best_deepC/best.exe [SUCCESS] Saved model as executable "best_deepC/best.exe"
There we have the model which can detect the hand-drawn shape using a neural network.
Notebook Link- Here
Credits- Siddharth Ganjoo
Also Read: Lower Back Pain Symptoms Detection using NN