if len(sys.argv) != 4: print('You need to input 4 arguments into the command line') '''=======================================IMPORT IMAGE DATA==========================================''' #passing data from command line testData = np.load(sys.argv[1]) #sys.argv[1] #the file to print generated labels will be sys.argv[2] #the decision on which model will be sys.argv[3] #testLabels = np.load(sys.argv[2]) '''=======================================PRE-PROCESSING IMAGE DATA==========================================''' for index in range(len(testData)): testData[index] = testData[index].astype(np.uint8) NormalizeImage(testData) Drawing_list = createContours(testData.shape[0], 'Training_images') '''=======================================DIAGONAL FEATURE EXTRACTION==========================================''' features = featureExtraction(Drawing_list) '''=======================================PREDICTING HANDWRITTEN CHARACTERS==========================================''' #Condition for selecting model if sys.argv[3] == 'extra': knn = joblib.load('knn_model_extra.joblib') else: knn = joblib.load('knn_model.joblib') test_predictions = knn.predict(features) #write test predictions to argv[2] outfile = sys.argv[2] + '.npy' print(test_predictions) np.save(outfile, np.asarray(test_predictions))
#trainData_AB = np.asarray(trainData_AB) #trainLabels_AB = np.asarray(trainLabels_AB) #pdb.set_trace() '''=========================================RESIZE AND SAVE AS BITMAP IMAGES========================================''' #NormalizeImage(trainData) NormalizeImage(trainData_1) NormalizeImage(trainData_2) pdb.set_trace() '''=========================================CONTOUR IMAGES========================================''' #Drawing_list = fc.createContours(trainData.shape[0],'Training_images') #features = featureExtraction(Drawing_list) Drawing_list_1 = fc.createContours(trainData_1.shape[0], 'Training_images') features_1 = featureExtraction(Drawing_list_1) Drawing_list_2 = fc.createContours(trainData_2.shape[0], 'Training_images') features_2 = featureExtraction(Drawing_list_2) #combined = np.concatenate((features,features_1),axis=0) #combined = np.concatenate((features_2,combined),axis=0) combined = np.concatenate((features_2, features_1), axis=0) pdb.set_trace() '''=======================================CLASSIFYING HANDWRITTEN CHARACTERS==========================================''' """ optimum_k = 3 optimum_test_size = 0.3 distance = 'euclidean' X_train,X_test,y_train,y_test = train_test_split(combined,combinedLabels,test_size=optimum_test_size,shuffle=True) predictions_final = classification(optimum_k,X_train,y_train,X_test,distance)