Upload breed-classification.ipynb

breed-classification.ipynb

@@ -0,0 +1 @@
+{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"# Let's import the necessary packages \nimport os \nimport numpy \n\nimport cv2  # preprocessing I need this \nimport warnings \nimport numpy as np\nimport pandas as pd \nimport matplotlib.pyplot as plt \nwarnings.filterwarnings('ignore')\n\nimport tensorflow as tf \nfrom keras.models import Sequential \nfrom tensorflow.keras.layers import Dense,Flatten,Conv2D,MaxPooling2D, GlobalAveragePooling2D\nfrom keras.preprocessing.image import ImageDataGenerator, load_img \n\nfrom IPython import display \ndisplay.set_matplotlib_formats('svg')","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","execution":{"iopub.status.busy":"2022-05-20T07:08:54.419848Z","iopub.execute_input":"2022-05-20T07:08:54.420244Z","iopub.status.idle":"2022-05-20T07:09:00.299011Z","shell.execute_reply.started":"2022-05-20T07:08:54.420131Z","shell.execute_reply":"2022-05-20T07:09:00.298201Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Data loading \ntrain_gen = ImageDataGenerator(rescale = 1./255) # 0 to 255 pixels \n\ntrain_data = train_gen.flow_from_directory('../input/breedclassification/Breed Classification', target_size = (224, 224), \n                                          batch_size = 32, class_mode = 'categorical', shuffle = True)","metadata":{"execution":{"iopub.status.busy":"2022-05-20T07:09:05.208729Z","iopub.execute_input":"2022-05-20T07:09:05.209183Z","iopub.status.idle":"2022-05-20T07:09:05.32511Z","shell.execute_reply.started":"2022-05-20T07:09:05.209148Z","shell.execute_reply":"2022-05-20T07:09:05.324383Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Transfer learning \nfrom keras.applications.mobilenet_v2 import MobileNetV2,preprocess_input \n\nmodel = Sequential()\nbase_model = MobileNetV2(input_shape = [224, 224, 3], weights = 'imagenet', include_top = False)  ","metadata":{"execution":{"iopub.status.busy":"2022-05-20T07:17:34.99469Z","iopub.execute_input":"2022-05-20T07:17:34.994957Z","iopub.status.idle":"2022-05-20T07:17:38.910577Z","shell.execute_reply.started":"2022-05-20T07:17:34.994929Z","shell.execute_reply":"2022-05-20T07:17:38.909782Z"},"collapsed":true,"jupyter":{"outputs_hidden":true},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# model correction \nbase_model.trainable = True  \n\nfor i in base_model.layers[:100]: \n    base_model.trainable = False\n    \npool = GlobalAveragePooling2D()\nmid_layer = Dense(100,activation = 'relu')\nfinal_1 = Dense(7,activation = 'softmax')\n\nmodel = Sequential([base_model, mid_layer,pool, final_1])\nmodel.summary()","metadata":{"execution":{"iopub.status.busy":"2022-05-20T07:20:46.220311Z","iopub.execute_input":"2022-05-20T07:20:46.220888Z","iopub.status.idle":"2022-05-20T07:20:47.030742Z","shell.execute_reply.started":"2022-05-20T07:20:46.220854Z","shell.execute_reply":"2022-05-20T07:20:47.030037Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"model.compile(loss = 'categorical_crossentropy',optimizer = 'adam',metrics = 'accuracy')\nmodel.fit(train_data, epochs = 100)  # one variable (history)","metadata":{"execution":{"iopub.status.busy":"2022-05-20T07:30:26.811674Z","iopub.execute_input":"2022-05-20T07:30:26.811964Z","iopub.status.idle":"2022-05-20T08:11:30.922475Z","shell.execute_reply.started":"2022-05-20T07:30:26.81193Z","shell.execute_reply":"2022-05-20T08:11:30.921801Z"},"collapsed":true,"jupyter":{"outputs_hidden":true},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Plot the accuracy! \n# plt.plot(history.history['acc'])\n# plt.title('model accuracy')\n# plt.ylabel('accuracy')\n# plt.xlabel('epoch')\n# plt.legend('train', loc='upper left')\n# plt.show()\n\n# # Plot the loss! \n# plt.plot(model.history['loss'])\n# plt.title('model loss')\n# plt.ylabel('loss')\n# plt.xlabel('epoch')\n# plt.legend('Loss', loc='upper left')\n# plt.show()","metadata":{"execution":{"iopub.status.busy":"2022-05-20T08:12:21.391741Z","iopub.execute_input":"2022-05-20T08:12:21.392578Z","iopub.status.idle":"2022-05-20T08:12:21.397194Z","shell.execute_reply.started":"2022-05-20T08:12:21.39253Z","shell.execute_reply":"2022-05-20T08:12:21.396297Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Predict the new thing \ndef predict_classes(link): \n    img = cv2.imread(link)\n    img = cv2.resize(img,(224,224))\n    img = img/255\n    img = img.reshape(-1,224,224,3)    # -1 is  for reshaping basically \n    pred = np.round(model.predict(img)).argmax(axis = 1)\n    dic = {0: 'Herding breed', 1: 'Hound breed', 2: 'Non sporting breed', 3: 'Terrior breed', 4:'working breed', 5: 'sporting breed', 6: 'toy breed'}\n    print(dic.get(int(pred)))","metadata":{"execution":{"iopub.status.busy":"2022-05-20T08:12:24.374276Z","iopub.execute_input":"2022-05-20T08:12:24.375045Z","iopub.status.idle":"2022-05-20T08:12:24.380949Z","shell.execute_reply.started":"2022-05-20T08:12:24.374999Z","shell.execute_reply":"2022-05-20T08:12:24.380296Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# predict \nlink = './sample_image.png'\npredict_classes(link)\nprint()\nload_img(link)","metadata":{"execution":{"iopub.status.busy":"2022-05-20T08:35:03.426257Z","iopub.execute_input":"2022-05-20T08:35:03.426523Z","iopub.status.idle":"2022-05-20T08:35:07.008956Z","shell.execute_reply.started":"2022-05-20T08:35:03.426494Z","shell.execute_reply":"2022-05-20T08:35:07.007834Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# classification report \ntrue_classes = train_data.classes\nclass_labels = list(train_data.class_indices.keys())","metadata":{"execution":{"iopub.status.busy":"2022-05-20T08:14:55.038042Z","iopub.execute_input":"2022-05-20T08:14:55.038798Z","iopub.status.idle":"2022-05-20T08:14:55.043095Z","shell.execute_reply.started":"2022-05-20T08:14:55.038756Z","shell.execute_reply":"2022-05-20T08:14:55.042264Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from sklearn.metrics import classification_report,plot_confusion_matrix\n\ny_hat = model.predict(train_data)\npredicted_classes = np.argmax(y_hat, axis = 1)\nreport = classification_report(true_classes, predicted_classes, target_names = class_labels)\nprint(report)","metadata":{"execution":{"iopub.status.busy":"2022-05-20T08:15:02.860499Z","iopub.execute_input":"2022-05-20T08:15:02.860763Z","iopub.status.idle":"2022-05-20T08:15:27.820247Z","shell.execute_reply.started":"2022-05-20T08:15:02.860734Z","shell.execute_reply":"2022-05-20T08:15:27.819427Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# saving the model \nmodel.save('breedclassification.h5')\n\n","metadata":{"execution":{"iopub.status.busy":"2022-05-20T08:27:44.88216Z","iopub.execute_input":"2022-05-20T08:27:44.882983Z","iopub.status.idle":"2022-05-20T08:27:45.324031Z","shell.execute_reply.started":"2022-05-20T08:27:44.882876Z","shell.execute_reply":"2022-05-20T08:27:45.323219Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# to load the model \n# new_model = tf.keras.models.load_model('breedclassification.h5')\n# new_model.predict(test_data) ","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import requests\nresponse = requests.get(\"https://www.groomers-online.com/images/cms/sections/1598536220-35522500.png\")\n\nfile = open(\"sample_image.png\", \"wb\")\nfile.write(response.content)\nfile.close()","metadata":{"execution":{"iopub.status.busy":"2022-05-20T08:43:50.454966Z","iopub.execute_input":"2022-05-20T08:43:50.455586Z","iopub.status.idle":"2022-05-20T08:43:51.882885Z","shell.execute_reply.started":"2022-05-20T08:43:50.455548Z","shell.execute_reply":"2022-05-20T08:43:51.882115Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# predict \nlink = './sample_image.png'\npredict_classes(link)\nprint()\nload_img(link)","metadata":{"execution":{"iopub.status.busy":"2022-05-20T08:44:00.376021Z","iopub.execute_input":"2022-05-20T08:44:00.376748Z","iopub.status.idle":"2022-05-20T08:44:00.522812Z","shell.execute_reply.started":"2022-05-20T08:44:00.376709Z","shell.execute_reply":"2022-05-20T08:44:00.522111Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# to load the model \nnew_model = tf.keras.models.load_model('breedclassification.h5')\n","metadata":{"execution":{"iopub.status.busy":"2022-05-20T08:44:51.233015Z","iopub.execute_input":"2022-05-20T08:44:51.233722Z","iopub.status.idle":"2022-05-20T08:44:52.648856Z","shell.execute_reply.started":"2022-05-20T08:44:51.233684Z","shell.execute_reply":"2022-05-20T08:44:52.648108Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Predict the new thing \ndef predict_classes(link): \n    img = cv2.imread(link)\n    img = cv2.resize(img,(224,224))\n    img = img/255\n    img = img.reshape(-1,224,224,3)    # -1 is  for reshaping basically \n    pred = np.round(new_model.predict(img)).argmax(axis = 1)\n    dic = {0: 'Herding breed', 1: 'Hound breed', 2: 'Non sporting breed', 3: 'Terrior breed', 4:'working breed', 5: 'sporting breed', 6: 'toy breed'}\n    print(dic.get(int(pred)))\n\n# predict \nlink = './sample_image.png'\npredict_classes(link)\nprint()\nload_img(link)","metadata":{"execution":{"iopub.status.busy":"2022-05-20T08:45:29.228947Z","iopub.execute_input":"2022-05-20T08:45:29.229628Z","iopub.status.idle":"2022-05-20T08:45:30.263623Z","shell.execute_reply.started":"2022-05-20T08:45:29.229588Z","shell.execute_reply":"2022-05-20T08:45:30.262901Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"! pip install gradio ","metadata":{"execution":{"iopub.status.busy":"2022-05-20T08:49:34.944264Z","iopub.execute_input":"2022-05-20T08:49:34.944615Z","iopub.status.idle":"2022-05-20T08:49:53.386843Z","shell.execute_reply.started":"2022-05-20T08:49:34.944576Z","shell.execute_reply":"2022-05-20T08:49:53.385866Z"},"collapsed":true,"jupyter":{"outputs_hidden":true},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import gradio as gr\nimage = gr.inputs.Image(shape=(224, 224))\nlabel = gr.outputs.Label(num_top_classes=7)\n\ngr.Interface(fn=predict_classes, inputs=image, outputs=label,interpretation='default').launch() ","metadata":{"execution":{"iopub.status.busy":"2022-05-20T08:50:55.632124Z","iopub.execute_input":"2022-05-20T08:50:55.632714Z","iopub.status.idle":"2022-05-20T08:50:59.347486Z","shell.execute_reply.started":"2022-05-20T08:50:55.632677Z","shell.execute_reply":"2022-05-20T08:50:59.346601Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}