Dev

.gitignore

@@ -0,0 +1,3 @@
+
+*.h5
+env

1_Image_Annotation/Convert_to_YOLO_format.py

@@ -0,0 +1,78 @@
+from PIL import Image
+from os import path, makedirs
+import os
+import re 
+import pandas as pd
+import sys
+import argparse
+def get_parent_dir(n=1):
+    """ returns the n-th parent dicrectory of the current
+    working directory """
+    current_path = os.path.dirname(os.path.abspath(__file__))
+    for k in range(n):
+        current_path = os.path.dirname(current_path)
+    return current_path
+
+sys.path.append(os.path.join(get_parent_dir(1),'Utils'))
+from Convert_Format import convert_vott_csv_to_yolo
+
+Data_Folder = os.path.join(get_parent_dir(1),'Data')
+VoTT_Folder = os.path.join(Data_Folder,'Source_Images','vott-csv-export')
+VoTT_csv =  os.path.join(VoTT_Folder,'Annotations-export.csv')
+YOLO_filename = os.path.join(VoTT_Folder,'data_train.txt')
+
+model_folder =  os.path.join(Data_Folder,'Model_Weights')
+classes_filename = os.path.join(model_folder,'data_classes.txt')
+
+AWS_path = '/home/ubuntu/TrainYourOwnYOLO/Data/Source_Images/vott-csv-export/'
+
+
+
+
+if __name__ == '__main__':
+    # surpress any inhereted default values
+    parser = argparse.ArgumentParser(argument_default=argparse.SUPPRESS)
+    '''
+    Command line options
+    '''
+    parser.add_argument(
+        "--VoTT_Folder", type=str, default=VoTT_Folder,
+        help = "absolute path to the exported files from the image tagging step with VoTT."
+    )
+
+    parser.add_argument(
+        "--VoTT_csv", type=str, default=VoTT_csv,
+        help = "absolute path to the *.csv file exported from VoTT. The default name is 'Houses-export.csv'."
+    )
+    parser.add_argument(
+        "--YOLO_filename", type=str, default=YOLO_filename,
+        help = "absolute path to the file where the annotations in YOLO format should be saved. The default name is 'data_train.txt' and is saved in the VoTT folder."
+    )
+
+    parser.add_argument(
+        '--AWS', default=False, action="store_true",
+        help='Enable this flag if you plan to train on AWS but did your pre-processing on a local machine.'
+    )
+
+    FLAGS = parser.parse_args()
+
+    #Prepare the dataset for YOLO
+    multi_df = pd.read_csv(FLAGS.VoTT_csv)
+    labels = multi_df['label'].unique()
+    labeldict = dict(zip(labels,range(len(labels))))
+    multi_df.drop_duplicates(subset=None, keep='first', inplace=True)
+    if FLAGS.AWS:
+        train_path = AWS_path
+    else:
+        train_path = FLAGS.VoTT_Folder
+    convert_vott_csv_to_yolo(multi_df,labeldict,path = train_path,target_name=FLAGS.YOLO_filename)
+
+    # Make classes file
+    file = open(classes_filename,"w") 
+    #Sort Dict by Values
+    SortedLabelDict = sorted(labeldict.items() ,  key=lambda x: x[1])
+    for elem in SortedLabelDict:
+	    file.write(elem[0]) 
+    file.close() 
+
+

1_Image_Annotation/README.md

@@ -0,0 +1,37 @@
+# TrainYourOwnYOLO: Image Annotation
+To train our YOLO detector, we first annotate images located in [`TrainYourOwnYOLO/Data/Source_Images`](/Data/Source_Images) with the help of Microsoft's Visual Object Tagging Tool (VoTT).
+
+## Download VoTT
+Head to VoTT [releases](https://github.com/Microsoft/VoTT/releases) and download and install the version for your operation system. 
+
+## Create a New Project
+
+![New Project](/1_Image_Annotation/Screenshots/New_Project.gif)
+
+Create a **New Project** and call it `Annotations`. It is highly recommeded to use `Annotations` as your project name. If you like to use a different name for your project, you will have to modify the command line arguments of subsequent scripts accordingly. 
+
+Under **Source Connection** choose **Add Connection** and put `Images` as **Display Name**. Under **Provider** choose **Local File System** and select the folder with [`Source Images`](/Data/Source_Images). For **Target Connection** choose the same folder as for **Source Connection**. Hit **Save Project** to finish project creation. 
+
+## Export Settings
+Navigate to **Export Settings** in the sidebar and then change the **Provider** to `Comma Seperated Values (CSV)`, then hit **Save Export Settings**. 
+
+![New Project](/1_Image_Annotation/Screenshots/Export_Settings.gif)
+
+
+## Labeling
+First create a new tag on the right and give it a relevant tag name. In our example, we choose `Cat_Face`. Then draw bounding boxes around your objects. You can use the number key to quickly assign tags to the current bounding box. 
+
+![New Project](/1_Image_Annotation/Screenshots/Labeling.gif)
+
+## Export Results
+Once you have labeled enough images (try to label at least 100 objects) press **CRTL+E** to export the project. You should now see a folder called `vott-csv-export` in the [`Source Images`](/Data/Source_Images) directory. Within that folder, you should see a `*.csv` file called `Annotations-export.csv` which contains information on all bounding boxes. 
+
+## Convert to YOLO Format
+As a final step, convert the VoTT csv format to the YOLOv3 format. To do so, run the conversion script:
+
+```
+python Convert_to_YOLO_format.py
+```
+The script generates two output files: `data_train.txt` located in the [`vott-csv-export`](/Data/Source_Images/vott-csv-export)) folder and `data_classes.txt` located in the [`TrainYourOwnYOLO/Data/Model_Weights`](/Data/Model_Weights/) folder.
+
+That's all for annotation! Next, go to [`TrainYourOwnYOLO/2_Training`](/2_Training) to train your YOLOv3 detector.

2_Training/AWS/README.md

@@ -0,0 +1,11 @@
+# TrainYourOwnYOLO: Training on AWS
+
+If your local machine does not have a GPU, training could take a very long time. To speed things up use an AWS GPU instance. 
+
+## Spinning up a GPU Instance
+To spin up a GPU instance, go to **EC2** and select **Launch Instance**. Then go to **Deep Learning AMI (Ubuntu) Version xx.x ** and hit **Select**. Under instance type, select **p2.xlarge** as the Instance Type. Proceed by hitting **Review and Launch**. 
+
+![Deep_Learning_AMI](/2_Training/AWS/Screenshots/AWS_Deep_Learning_AMI.gif)
+
+## Start the Training
+Connect to your instance and follow the same steps as on your local machine. Make sure that all your Source Images are in [`TrainYourOwnYOLO/Data/Source_Images`](/Data/Source_Images) and that the [`data_train.txt`](/Data/Source_Images/vott-csv-export/data_train.txt) and [`data_classes.txt`](/Data/Model_Weights/data_classes.txt) are up-to-date.

2_Training/Download_YOLO_weights.py

@@ -0,0 +1,84 @@
+import os
+import subprocess
+import time
+import sys
+
+# TrainYourOwnYOLO/Training/src/keras_yolo3
+
+root_folder = os.path.dirname(os.path.abspath(__file__))
+
+
+download_folder = os.path.join(root_folder,'src','keras_yolo3')
+
+convert_script = os.path.join(download_folder,'Download_YOLO_weights.sh')
+
+call_string = 'bash ' + convert_script
+
+subprocess.call(call_string , shell=True)
+
+
+# weight_file = os.path.join(download_folder,'yolov3.weights')
+# print(download_folder)
+# # wget https://pjreddie.com/media/files/yolov3.weights
+# # python convert.py yolov3.cfg yolov3.weights model_data/yolo.h5
+
+# # call_string = 'wget https://pjreddie.com/media/files/yolov3.weights -O ' + weight_file
+
+# # subprocess.call(call_string , shell=True)
+
+
+# # wget http://download.files.com/software/files.tar.gz -O /home/yourname/Downloads
+
+# convert_script = weight_file = os.path.join(download_folder,'convert.py')
+
+# call_string = 'python ' + convert_script + ' yolov3.cfg yolov3.weights model_data/yolo.h5'
+# subprocess.call(call_string , shell=True)
+
+# # python convert.py yolov3.cfg yolov3.weights model_data/yolo.h5
+
+
+# def make_call_string(arglist):
+#     result_string = ''
+#     for arg in arglist:
+#         result_string+= ''.join(['--',arg[0],' ', arg[1],' '])
+#     return result_string
+# root_folder = os.path.dirname(os.path.abspath(__file__))
+# data_folder = os.path.join(root_folder,'Data')
+# model_folder = os.path.join(data_folder,'Model_Weights')
+# image_folder = os.path.join(data_folder,'Source_Images')
+# input_folder =  os.path.join(image_folder,'Test_Images')
+# output_folder =  os.path.join(image_folder,'Test_Image_Detection_Results')
+
+
+# if not os.path.exists(output_folder):
+#     os.mkdir(output_folder)
+
+# #First download the pre-trained weights
+# download_script = os.path.join(model_folder,'Download_Weights.py')
+
+# print('Downloading Pretrained Weights')
+# start = time.time()
+# call_string = ' '.join(['python',download_script,'1MGXAP_XD_w4OExPP10UHsejWrMww8Tu7', os.path.join(model_folder,'trained_weights_final.h5') ])
+
+# subprocess.call(call_string , shell=True)
+
+# end = time.time()
+# print('Downloaded Pretrained Weights in {0:.1f} seconds'.format(end-start))
+
+# # Now run the cat face detector
+# detector_script = os.path.join(os.path.dirname(os.path.abspath(__file__)),'3_Inference','Detector.py')
+
+
+# result_file =  os.path.join(output_folder, 'Cat_Faces_Results.csv')
+# model_weights = os.path.join(model_folder,'trained_weights_final.h5')
+# classes_file = os.path.join(model_folder,'data_classes.txt')
+# anchors = os.path.join(root_folder,'2_Training','src','keras_yolo3','model_data','yolo_anchors.txt')
+
+# arglist = [['input_images',input_folder],['classes',classes_file],['output',output_folder],['yolo_model',model_weights],['box_file',result_file],['anchors',anchors]]
+# call_string = ' '.join(['python', detector_script,make_call_string(arglist)])
+
+# print('Detecting Cat Faces by calling \n\n', call_string,'\n')
+# start = time.time()
+# subprocess.call(call_string, shell=True)
+# end = time.time()
+# print('Detected Cat Faces in {0:.1f} seconds'.format(end-start))

2_Training/README.md

@@ -0,0 +1,20 @@
+# TrainYourOwnYOLO: Training
+Using the training images located in [`TrainYourOwnYOLO/Data/Source_Images`](/Data/Source_Images) together with our annotation file [`data_train.txt`](/Data/Source_Images/vott-csv-export) we can train our YOLOv3 detector. 
+
+## Download the Pre-Trained Weights
+Before getting started we need to download the pre-trained YOLOv3 weights and convert them to the keras format. To run both steps execute:
+
+```
+python Download_YOLO_weights.py
+```
+## Train YOLOv3 Detector
+To start the training run the training script:
+```
+python Train_YOLO.py 
+```
+Depending on your set-up this process can take a few minutes to a few hours. I recommend using a GPU to speed up training. The final weights are saved in [`TrainYourOwnYOLO/Data/Model_weights`](/Data/Model_weights). 
+
+### Training on AWS
+If training is too slow on your local machine, you can speed up the training by using a GPU cluster on AWS. To learn more about training on AWS navigate to the [`AWS`](/2_Training/AWS) folder.
+
+This concludes the training step and you are now ready to detect cat faces in new images!