YOLOv3_Custom_tomato_ripneness_.ipynb

{"cells":[{"cell_type":"markdown","metadata":{"id":"mZVUXbd1RgSb"},"source":["\n","# **Step 1**\n","## Enabling and testing the GPU\n","\n","The notebook's GPUs must first be enabled:\n","\n","- Select Notebook Settings under Edit.\n","- choose GPU using the Hardware Accelerator drop-down\n","\n","Next, we'll check if Tensorflow can connect to the GPU: By executing the following code, you may quickly determine whether the GPU is enabled."]},{"cell_type":"code","execution_count":null,"metadata":{"id":"JLpDKNVgH-xd"},"outputs":[],"source":["# Check if NVIDIA GPU is enabled\n","!nvidia-smi"]},{"cell_type":"markdown","metadata":{"id":"X_aBTDRnGXSB"},"source":["# **Step 2**\n","Mounting the Drive to store and load files."]},{"cell_type":"code","execution_count":null,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":21423,"status":"ok","timestamp":1697185544233,"user":{"displayName":"Daniel Yang","userId":"08998102689171284536"},"user_tz":-780},"id":"YmNIlZIMapG0","outputId":"c676647e-546a-43a2-f0e2-7187ba9c0920"},"outputs":[{"name":"stdout","output_type":"stream","text":["Mounted at /content/gdrive\n"]}],"source":["from google.colab import drive\n","drive.mount('/content/gdrive')"]},{"cell_type":"markdown","metadata":{"id":"Zd7NjfKC6qxX"},"source":["Let's make the required files and directories for training with custom objects.\n","\n","\n","1. **custom_data/images**\n","2. **custom_data/custom.names**\n","3. **custom_data/train.txt**\n","4. **custom_data/test.txt**\n","5. **custom_data/backup**\n","6. **custom_data/detector.data**\n","7. **custom_data/cfg**"]},{"cell_type":"markdown","metadata":{"id":"yWGXMV9M9iJB"},"source":["# **Step 3**\n","## *Creating Image Directory*\n","We will train our model to recognise pistols in this project, thus we must gather the images and its annotaions and save them in the ***custom data/images*** directory."]},{"cell_type":"markdown","metadata":{"id":"YaPD3ZxdI8Cy"},"source":["## Unzip Files\n","* Your Google Drive file location that you wish to unzip is the first path in the cell below.\n","* Where you wish to extract the zip file from Google Drive as shown by the second path in the cell below.\n","\n","\n","In our case text files should be saved in **custom_data/images** directory. For e.g. **image1.jpg** should have a text file **image1.txt**."]},{"cell_type":"code","execution_count":null,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":4605,"status":"ok","timestamp":1697068501093,"user":{"displayName":"Daniel Yang","userId":"08998102689171284536"},"user_tz":-780},"id":"GKbY7BsgH943","outputId":"08311a33-480c-478d-d30b-22e6be55046f"},"outputs":[{"name":"stdout","output_type":"stream","text":["Archive:  /content/gdrive/MyDrive/custom_data/zip/images.zip\n","replace /content/gdrive/MyDrive/custom_data/images/008812fa2df9d65fae04ce6c14a3e54e.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: "]}],"source":["!unzip \"/content/gdrive/MyDrive/custom_data/zip/images.zip\" -d \"/content/gdrive/MyDrive/custom_data/images\""]},{"cell_type":"markdown","metadata":{"id":"sUyoW6Be09T1"},"source":["# **Step 4**\n","## *Creating Custom.names file*\n","Labels of our objects should be saved in **custom_data/custom.names** file, each line in the file corresponds to an object. In our case since we have only one object class, the file should contain the following.\n"]},{"cell_type":"markdown","metadata":{"id":"MCfztxWBTy8j"},"source":["# **Step 5**\n","## *Creating Train and Test files*\n","The annotated photos can then be randomly split into train and test sets in a **80:20** ratio.\n","\n","**custom data/train.txt** The location of the train dataset should be listed in each file row.\n","\n","**custom data/test.txt** The location of the test dataset should be listed in each file row.\n"]},{"cell_type":"markdown","metadata":{"id":"Aq_8H9TpF6mh"},"source":["# **Step 6**\n","## *Creating Backup directory*\n","Creating backup directory for storing weights of the trained model."]},{"cell_type":"markdown","metadata":{"id":"98u2UBFwVhqo"},"source":["# **Step 7**\n","## *Creating YOLO data file*\n","Make a file called \"detector.data\" in the \"custom data\" directory that contains details about the train and test data sets."]},{"cell_type":"markdown","metadata":{"id":"bt-lz7YbV543"},"source":["![Screenshot from 2022-10-19 19-49-20.png](data:image/png;base64,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)"]},{"cell_type":"markdown","metadata":{"id":"e8MhpWw4as9X"},"source":["# **Step 8**\n","## *Cloning Directory to use Darknet*\n","Darknet, an open source neural network framework, will be used to train the detector. Download and create a dark network"]},{"cell_type":"code","execution_count":null,"metadata":{"id":"Dt-S5cL8Zgv2"},"outputs":[],"source":["!git clone https://github.com/AlexeyAB/darknet"]},{"cell_type":"code","execution_count":null,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":512,"status":"ok","timestamp":1697185558363,"user":{"displayName":"Daniel Yang","userId":"08998102689171284536"},"user_tz":-780},"id":"QogAehWGfKC5","outputId":"1329d750-b411-4725-a5ca-0ef8302d0c73"},"outputs":[{"name":"stdout","output_type":"stream","text":["/content/darknet\n"]}],"source":["# Change current working directory to Darknet\n","%cd darknet"]},{"cell_type":"markdown","metadata":{"id":"s5iGbzq-UFZT"},"source":["### Change makefile to have GPU and OPENCV enabled, and other parameters for faster computation."]},{"cell_type":"code","execution_count":null,"metadata":{"id":"PZzl5qJvZiCF"},"outputs":[],"source":["!sed -i 's/OPENCV=0/OPENCV=1/' Makefile\n","!sed -i 's/GPU=0/GPU=1/' Makefile\n","!sed -i 's/CUDNN=0/CUDNN=1/' Makefile\n","!sed -i 's/CUDNN_HALF=0/CUDNN_HALF=1/' Makefile"]},{"cell_type":"code","execution_count":null,"metadata":{"id":"QHncrMa623qR"},"outputs":[],"source":["# Make file\n","\"/content/darknet/Makefile\""]},{"cell_type":"markdown","metadata":{"id":"xWKWcUQCUyj2"},"source":["### Model Compilation"]},{"cell_type":"code","execution_count":null,"metadata":{"id":"J-R3QHDiUwpf"},"outputs":[],"source":["# compiling the model\n","!make"]},{"cell_type":"markdown","metadata":{"id":"gBl4eHGrWgVI"},"source":["# **Step 9**\n","## *Making changes in the yolo Configuration file*\n"]},{"cell_type":"markdown","metadata":{"id":"nwTXwdCsTei2"},"source":["# **Step 10**\n","## *Downloading Pre-trained weights*\n","To train our object detector, we can use the pre-trained weights that have already been trained on a large data sets. The pre-trained weights are available [here](https://pjreddie.com/media/files/darknet53.conv.74), and they can be downloaded to the root directory."]},{"cell_type":"code","execution_count":null,"metadata":{"id":"aihq0RvLvsWa"},"outputs":[],"source":["# Download weights darknet model 53\n","!wget https://pjreddie.com/media/files/darknet53.conv.74"]},{"cell_type":"code","execution_count":null,"metadata":{"id":"WF0QSFMfAeqG"},"outputs":[],"source":["# define helper functions\n","def imShow(path):\n","  import cv2\n","  import matplotlib.pyplot as plt\n","  %matplotlib inline\n","\n","  image = cv2.imread(path)\n","  height, width = image.shape[:2]\n","  resized_image = cv2.resize(image,(3*width, 3*height), interpolation = cv2.INTER_CUBIC)\n","\n","  fig = plt.gcf()\n","  fig.set_size_inches(18, 10)\n","  plt.axis(\"off\")\n","  plt.imshow(cv2.cvtColor(resized_image, cv2.COLOR_BGR2RGB))\n","  plt.show()\n"]},{"cell_type":"markdown","metadata":{"id":"WJDvy-sXwGs3"},"source":["# **Step 11**\n","## *Training the model*\n","As soon as we have all the necessary files and annotated photographs, we can begin our training.\n","Up till the loss reaches a predefined level, we can keep training. Weights for the custom detector are initially saved once every 100 iterations until 1,000 iterations, after which they are saved once every 10,000 iterations by default.\n","\n","We can do detection using the generated weights after the training is finished."]},{"cell_type":"markdown","metadata":{"id":"ba-BaACTJyfQ"},"source":["## *Continue training from where you left*\n","Continue training from where you left off, your Model training can be stopped due to multiple reasons, like the notebook time out, notebook craches, due to network issues,  and many more,  so you can start your training from where you left off, by passing the previous trained weights."]},{"cell_type":"code","execution_count":null,"metadata":{"id":"xvUZgMX0wIz7"},"outputs":[],"source":["!./darknet detector train /content/gdrive/MyDrive/custom_data/detector.data /content/gdrive/MyDrive/custom_data/cfg/yolov3-custom.cfg darknet53.conv.74 -dont_show"]},{"cell_type":"code","execution_count":null,"metadata":{"id":"nBm-fr1n90x3"},"outputs":[],"source":["imShow('chart.png')"]},{"cell_type":"code","execution_count":null,"metadata":{"id":"BVkkRPlBJhAn"},"outputs":[],"source":["!./darknet detector train /content/gdrive/MyDrive/custom_data/detector.data /content/gdrive/MyDrive/custom_data/cfg/yolov3-custom.cfg /content/gdrive/MyDrive/custom_data/backup/yolov3-custom_final.weights -dont_show"]},{"cell_type":"code","execution_count":null,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":571,"status":"ok","timestamp":1697186591450,"user":{"displayName":"Daniel Yang","userId":"08998102689171284536"},"user_tz":-780},"id":"mpPD0SEwA0Yw","outputId":"23097820-a0ee-4894-a90d-46ca7a62313e"},"outputs":[{"name":"stdout","output_type":"stream","text":["/content/darknet/cfg\n","/content/darknet\n"]}],"source":["%cd cfg\n","!sed -i 's/batch=64/batch=1/' /content/gdrive/MyDrive/custom_data/cfg/yolov3-custom.cfg\n","!sed -i 's/subdivisions=16/subdivisions=1/' /content/gdrive/MyDrive/custom_data/cfg/yolov3-custom.cfg\n","%cd .."]},{"cell_type":"code","execution_count":null,"metadata":{"colab":{"background_save":true},"id":"mAr8mwtz_CPg"},"outputs":[],"source":["!./darknet detector test /content/gdrive/MyDrive/custom_data/detector.data /content/gdrive/MyDrive/custom_data/cfg/yolov3-custom.cfg /content/gdrive/MyDrive/custom_data/backup/yolov3-custom_final.weights /content/gdrive/MyDrive/custom_data/images/0233fb25d24d36370422d41086b52267.jpg -thresh 0.3\n","imShow('predictions.jpg')\n"]},{"cell_type":"code","execution_count":null,"metadata":{"id":"sbIUIVrNnBcq"},"outputs":[],"source":["!./darknet detector test /content/gdrive/MyDrive/custom_data/detector.data /content/gdrive/MyDrive/custom_data/cfg/yolov3-custom.cfg /content/gdrive/MyDrive/custom_data/backup/yolov3-custom_final.weights /content/gdrive/MyDrive/custom_data/images/tomaat2.jpg -thresh 0.3\n","imShow('predictions.jpg')\n"]},{"cell_type":"code","execution_count":null,"metadata":{"id":"5agajb9Tp-tE"},"outputs":[],"source":["!./darknet detector test /content/gdrive/MyDrive/custom_data/detector.data /content/gdrive/MyDrive/custom_data/cfg/yolov3-custom.cfg /content/gdrive/MyDrive/custom_data/backup/yolov3-custom_final.weights /content/gdrive/MyDrive/custom_data/images/IMG_2308.jpg -thresh 0.3\n","imShow('predictions.jpg')\n"]},{"cell_type":"code","source":["!./darknet detector test /content/gdrive/MyDrive/custom_data/detector.data /content/gdrive/MyDrive/custom_data/cfg/yolov3-custom.cfg /content/gdrive/MyDrive/custom_data/backup/yolov3-custom_final.weights /content/gdrive/MyDrive/custom_data/images/unnamed.jpg -thresh 0.3\n","imShow('predictions.jpg')"],"metadata":{"id":"5kghd_tVO0YI"},"execution_count":null,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"yLeZ-tcLEdS5"},"source":["# **Step 12**\n","## *Calculating Mean average precision of Specific Weights*\n"]},{"cell_type":"code","execution_count":null,"metadata":{"id":"RFLc7ablDccK"},"outputs":[],"source":["!./darknet detector map /content/gdrive/MyDrive/custom_data/detector.data /content/gdrive/MyDrive/custom_data/cfg/yolov3-custom.cfg /content/gdrive/MyDrive/custom_data/backup/yolov3-custom_final.weights -dont_show"]}],"metadata":{"accelerator":"GPU","colab":{"machine_shape":"hm","provenance":[]},"kernelspec":{"display_name":"Python 3","name":"python3"},"language_info":{"name":"python"}},"nbformat":4,"nbformat_minor":0}