Data_Augmentation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "ddb6d6d8",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from vidaug import augmentors as va\n",
    "import matplotlib.pyplot as plt\n",
    "import cv2\n",
    "import os\n",
    "import random\n",
    "from pathlib import Path"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "f09d684c",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Set up file location\n",
    "RWF_dataset_dir = \"/home/amrmarey/Work/ECE740_Project/RWF_2000\"  #location of original dataset\n",
    "RWF_sets = ['train']\n",
    "RWF_output_classes = ['Fight', 'NonFight']\n",
    "RWF_Aug_dir = \"/home/amrmarey/Work/ECE740_Project/RWF_2000_Flipped\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "8f9e518e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/home/amrmarey/Work/ECE740_Project/RWF_2000/train/Fight\n",
      "/home/amrmarey/Work/ECE740_Project/RWF_2000/train/NonFight\n",
      "Done!!!!!!!!\n"
     ]
    }
   ],
   "source": [
    "####This is for just flipping\n",
    "for RWF_set in RWF_sets: #Augment both train and val\n",
    "    for RWF_output_class in RWF_output_classes: #Augment each class Fight and NonFight\n",
    "        folder_path = RWF_dataset_dir +'/' + RWF_set + '/' + RWF_output_class\n",
    "        print(folder_path)\n",
    "        # Create a Path object for the folder\n",
    "        folder = Path(folder_path)\n",
    "        \n",
    "        # Get a list of all files in the folder\n",
    "        file_list = [file for file in folder.iterdir() if file.is_file()]\n",
    "        out_dir = RWF_Aug_dir + '/' + RWF_set + '/' + RWF_output_class\n",
    "        for file in file_list:\n",
    "            #Location of original file in computer\n",
    "            video_path = RWF_dataset_dir +'/' + RWF_set+ '/'+ RWF_output_class+ '/' + str(file).split(\"/\")[-1]\n",
    "            \n",
    "            #Location of wher you would like the output file to be\n",
    "            out_path = out_dir + '/' + str(file).split(\"/\")[-1][:-4] + '_flipped.avi'\n",
    "            # Load the video file\n",
    "            cap = cv2.VideoCapture(video_path)\n",
    "            # Define the output video writer\n",
    "            # Meta.\n",
    "            fps = int(cap.get(cv2.CAP_PROP_FPS))\n",
    "            width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))\n",
    "            height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))\n",
    "            frame_size = (width, height)\n",
    "            fourcc = cv2.VideoWriter_fourcc(*'mp4v')\n",
    "            # Video writer.\n",
    "            out = cv2.VideoWriter(out_path, fourcc, fps, frame_size)\n",
    "\n",
    "            # collect all frames of the video\n",
    "            frames = []\n",
    "            while(cap.isOpened()):\n",
    "                ret, frame = cap.read()\n",
    "                if ret == True:\n",
    "                    frames.append(frame)\n",
    "                else:\n",
    "                    break\n",
    "            cap.release()\n",
    "            #Data augmentations to consider\n",
    "            seq = va.Sequential([va.HorizontalFlip()])\n",
    "            #augment the frames\n",
    "            video_aug = seq(frames)\n",
    "\n",
    "            # output the video\n",
    "            for frame in video_aug:\n",
    "                out.write(frame)\n",
    "            out.release()\n",
    "\n",
    "            \n",
    "print('Done!!!!!!!!')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "c6a1614b",
   "metadata": {},
   "outputs": [],
   "source": [
    "#The rest of the script is responsible for noise injection\n",
    "RWF_dataset_dir = \"/home/amrmarey/Work/ECE740_Project/RWF_2000\" #location of original dataset\n",
    "RWF_sets = ['train']\n",
    "RWF_output_classes = ['Fight', 'NonFight']\n",
    "RWF_Aug_dir_v2 = \"/home/amrmarey/Work/ECE740_Project/RWF_2000_Aug\" #location of augmented dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "5b975292",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2\n",
      "3\n",
      "Done!!!!!!!!\n"
     ]
    }
   ],
   "source": [
    "#Noise injection for nonflipped videos\n",
    "M = 1\n",
    "while M < 3: #Run through the dataset twice\n",
    "    for RWF_set in RWF_sets:\n",
    "        for RWF_output_class in RWF_output_classes:\n",
    "            folder_path = RWF_dataset_dir +'/' + RWF_set + '/' + RWF_output_class\n",
    "            # Create a Path object for the folder\n",
    "            folder = Path(folder_path)\n",
    "\n",
    "            # Get a list of all files in the folder\n",
    "            file_list = [file for file in folder.iterdir() if file.is_file()]\n",
    "            out_dir = RWF_Aug_dir_v2 + '/' + RWF_set + '/' + RWF_output_class\n",
    "            for file in file_list:\n",
    "                #Location of original file in computer\n",
    "                video_path = RWF_dataset_dir +'/' + RWF_set+ '/'+ RWF_output_class+ '/' + str(file).split(\"/\")[-1]\n",
    "                \n",
    "                #Location of wher you would like the output file to be\n",
    "                out_path = out_dir + '/' + str(file).split(\"/\")[-1][:-4] + '_'+ str(M)+'.avi'\n",
    "                # Load the video file\n",
    "                cap = cv2.VideoCapture(video_path)\n",
    "                # Define the output video writer\n",
    "                # Meta.\n",
    "                fps = int(cap.get(cv2.CAP_PROP_FPS))\n",
    "                width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))\n",
    "                height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))\n",
    "                frame_size = (width, height)\n",
    "                fourcc = cv2.VideoWriter_fourcc(*'mp4v')\n",
    "                # Video writer.\n",
    "                out = cv2.VideoWriter(out_path, fourcc, fps, frame_size)\n",
    "\n",
    "                # collect all frames of the video\n",
    "                frames = []\n",
    "                while(cap.isOpened()):\n",
    "                    ret, frame = cap.read()\n",
    "                    if ret == True:\n",
    "                        frames.append(frame)\n",
    "                    else:\n",
    "                        break\n",
    "                cap.release()\n",
    "                # Apply the video augmentation pipeline to each frame of the video\n",
    "                \n",
    "                #Different Augmentations considered. Each video is randomly augmented with different augmentation and each augmentation has a random intensity\n",
    "                salt_pepper_ratio = random.randint(5, 90)\n",
    "                salt_and_pepper = va.Sequential(transforms = [(va.Salt(ratio = salt_pepper_ratio)),(va.Pepper(ratio = salt_pepper_ratio))], random_order = True)\n",
    "                seq = va.SomeOf(transforms = [va.GaussianBlur(sigma = random.uniform(1, 8)), salt_and_pepper, va.Multiply(value=random.uniform(0.85, 1.15))], N = random.randint(1, 3), random_order=True)\n",
    "\n",
    "                #augment the frames\n",
    "                video_aug = seq(frames)\n",
    "\n",
    "                # output the video\n",
    "                for frame in video_aug:\n",
    "                    out.write(frame)\n",
    "                out.release()\n",
    "            \n",
    "    M = M+1\n",
    "    print(M)\n",
    "\n",
    "            \n",
    "print('Done!!!!!!!!')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "78ac3331",
   "metadata": {},
   "outputs": [],
   "source": [
    "RWF_dataset_dir = \"/home/amrmarey/Work/ECE740_Project/RWF_2000_Flipped\"\n",
    "RWF_sets = ['train']\n",
    "RWF_output_classes = ['Fight', 'NonFight']\n",
    "RWF_Aug_dir_v2 = \"/home/amrmarey/Work/ECE740_Project/RWF_2000_Aug\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "f6648fc8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2\n",
      "3\n",
      "Done!!!!!!!!\n"
     ]
    }
   ],
   "source": [
    "#Noise injection for flipped videos (everything is same as above except folder locations change)\n",
    "M = 1\n",
    "while M < 3: #Run through the dataset twice\n",
    "    for RWF_set in RWF_sets:\n",
    "        for RWF_output_class in RWF_output_classes:\n",
    "            folder_path = RWF_dataset_dir +'/' + RWF_set + '/' + RWF_output_class\n",
    "            # Create a Path object for the folder\n",
    "            folder = Path(folder_path)\n",
    "\n",
    "            # Get a list of all files in the folder\n",
    "            file_list = [file for file in folder.iterdir() if file.is_file()]\n",
    "            out_dir = RWF_Aug_dir_v2 + '/' + RWF_set + '/' + RWF_output_class\n",
    "            for file in file_list:\n",
    "                #Location of original file in computer\n",
    "                video_path = RWF_dataset_dir +'/' + RWF_set+ '/'+ RWF_output_class+ '/' + str(file).split(\"/\")[-1]\n",
    "                \n",
    "                #Location of wher you would like the output file to be\n",
    "                out_path = out_dir + '/' + str(file).split(\"/\")[-1][:-4] + '_'+ str(M)+'.avi'\n",
    "                # Load the video file\n",
    "                cap = cv2.VideoCapture(video_path)\n",
    "                # Define the output video writer\n",
    "                # Meta.\n",
    "                fps = int(cap.get(cv2.CAP_PROP_FPS))\n",
    "                width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))\n",
    "                height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))\n",
    "                frame_size = (width, height)\n",
    "                fourcc = cv2.VideoWriter_fourcc(*'mp4v')\n",
    "                # Video writer.\n",
    "                out = cv2.VideoWriter(out_path, fourcc, fps, frame_size)\n",
    "\n",
    "                # collect all frames of the video\n",
    "                frames = []\n",
    "                while(cap.isOpened()):\n",
    "                    ret, frame = cap.read()\n",
    "                    if ret == True:\n",
    "                        frames.append(frame)\n",
    "                    else:\n",
    "                        break\n",
    "                cap.release()\n",
    "                # Apply the video augmentation pipeline to each frame of the video\n",
    "                \n",
    "                #Different Augmentations considered. Each video is randomly augmented with different augmentation and each augmentation has a random intensity\n",
    "                salt_pepper_ratio = random.randint(5, 90)\n",
    "                salt_and_pepper = va.Sequential(transforms = [(va.Salt(ratio = salt_pepper_ratio)),(va.Pepper(ratio = salt_pepper_ratio))], random_order = True)\n",
    "                seq = va.SomeOf(transforms = [va.GaussianBlur(sigma = random.uniform(1, 8)), salt_and_pepper, va.Multiply(value=random.uniform(0.85, 1.15))], N = random.randint(1, 3), random_order=True)\n",
    "\n",
    "                #augment the frames\n",
    "                video_aug = seq(frames)\n",
    "\n",
    "                # output the video\n",
    "                for frame in video_aug:\n",
    "                    out.write(frame)\n",
    "                out.release()\n",
    "            \n",
    "    M = M+1\n",
    "    print(M)\n",
    "\n",
    "            \n",
    "print('Done!!!!!!!!')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}