IoU_Metric.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "authorship_tag": "ABX9TyPwkMCRl36Ap5CvaaUyGOYQ",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/sineeli/GANs-JAX/blob/main/IoU_Metric.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Implement IoU using JAX"
      ],
      "metadata": {
        "id": "nmYzKldj7abi"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "id": "xMyh0jGD9Paf"
      },
      "outputs": [],
      "source": [
        "import cv2\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "import jax\n",
        "import jax.numpy as jnp"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_two_bbox(image, gt_bbox, pd_bbox):\n",
        "  # Draw the box with desired color and thickness but without filling\n",
        "  cv2.rectangle(image, (gt_bbox[0], gt_bbox[1]),\n",
        "  (gt_bbox[0] + gt_bbox[2], gt_bbox[1] + gt_bbox[3]),\n",
        "    (0, 255, 0), 1)  # Green color, thickness 1\n",
        "\n",
        "  cv2.rectangle(image, (pd_bbox[0], pd_bbox[1]),\n",
        "  (pd_bbox[0] + pd_bbox[2], pd_bbox[1] + pd_bbox[3]),\n",
        "    (255, 0, 0), 1)  # Red color, thickness 1\n",
        "\n",
        "  # Display the image\n",
        "  plt.imshow(image)\n",
        "  cv2.waitKey(0)\n",
        "  plt.axis(\"off\")\n",
        "  cv2.destroyAllWindows()"
      ],
      "metadata": {
        "id": "RQkbbxNR3aO1"
      },
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Define image (you can replace this with your image loading code)\n",
        "image = np.ones((256, 256, 3), dtype=np.uint8) * 255  # Create a white image"
      ],
      "metadata": {
        "id": "BSTF3wCZ5Jap"
      },
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Draw different possibilities of boxes"
      ],
      "metadata": {
        "id": "oThuJIKo7R0P"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Slight Overlap"
      ],
      "metadata": {
        "id": "SpyeR9d549nw"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "plot_two_bbox(image.copy(), [50, 72, 197, 121], [20, 20, 90, 120])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "DyHCy-eG0gMZ",
        "outputId": "55fa6b2f-a6fd-44b3-9704-9946928ba0a3"
      },
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### No Overlap"
      ],
      "metadata": {
        "id": "SJ5MZN6e5ECI"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "plot_two_bbox(image.copy(), [39, 63, 203, 112], [10, 10, 40, 40])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "tE8tIgj55ARg",
        "outputId": "ed119e69-3672-4c3c-d3c3-9ebb50173d6d"
      },
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Max Overlap"
      ],
      "metadata": {
        "id": "i4irI9Dw5VQ_"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "plot_two_bbox(image.copy(), [49, 75, 203, 125], [42, 78, 186, 126])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "H6bXnnrF5V3R",
        "outputId": "c5f2988e-80b2-434b-ac61-417e1cc2f002"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Box inside Box"
      ],
      "metadata": {
        "id": "xrPm7Xbf5g_B"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "plot_two_bbox(image.copy(), [31, 69, 201, 125], [18, 63, 235, 135])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "T9Tpr3xa5a2B",
        "outputId": "a86e0db0-aca9-4c65-df0c-da0855dad7f8"
      },
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Max Overlap 2"
      ],
      "metadata": {
        "id": "3hKPVNoB5plg"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "plot_two_bbox(image.copy(), [50, 72, 197, 121], [54, 72, 198, 120])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "tjnhV-VB5jAY",
        "outputId": "b1ffae86-4b4d-4a7f-b95a-0b5a21b4d7a1"
      },
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Intersection Over Union(IoU)\n",
        "\n",
        "$$ IoU = \\dfrac{area \\, of \\, overlap}{area\\ of\\ union} $$\n",
        "\n",
        "<img src=\"https://machinelearningspace.com/wp-content/uploads/2023/01/IOU2.jpg\"/>\n",
        "\n",
        "\n",
        "Okay, let's now breakdown the problem and solve it step by step.\n",
        "* We need intersection area of the two bounding boxes.\n",
        "* $Area \\,\\,of \\,\\,Overlap → A\\cap B$ (Numerator):\n",
        "  * Taking `maxmimum` of `x-coordinates` and `y-coordinates` which is basically top left corner of the intersection.\n",
        "  * Taking `minumum` of `x-coordinates` and `y-coordinates`which is basically bottom right corner of the intersection.\n",
        "* $Area \\,\\,of \\,\\,Union → A\\cup B$ (Denominator):\n",
        "  * $Area \\,of \\,BBox1 + Area \\,of \\,BBox2 - A \\cap B$\n",
        "\n",
        "* Now if intersection fo boxes `zero` would be a edge case in that scenario values may become less than zero values.\n",
        "\n",
        "\n",
        "* [img-ref](https://machinelearningspace.com/intersection-over-union-iou-a-comprehensive-guide/)"
      ],
      "metadata": {
        "id": "RKEP7E9r7GwM"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from typing import Union, List, Any, NoReturn\n",
        "import numpy as jnp\n",
        "\n",
        "def iou(gt_boxes: Union[List[int], jnp.ndarray], boxes: Union[List[int], jnp.ndarray]) -> jnp.ndarray:\n",
        "  \"\"\"\n",
        "  Calculates the Intersection over Union (IoU) between ground truth boxes and predicted boxes.\n",
        "\n",
        "  Args:\n",
        "      gt_boxes: A list of integers or a NumPy array of shape (num_boxes, 4) representing the ground truth bounding boxes.\n",
        "                 Each box is represented by top-left (x1, y1) and bottom-right (x2, y2) coordinates.\n",
        "      boxes: A list of integers or a NumPy array of shape (num_boxes, 4) representing the predicted bounding boxes.\n",
        "             Each box is represented by top-left (x1, y1) and bottom-right (x2, y2) coordinates.\n",
        "\n",
        "  Returns:\n",
        "      A NumPy array of shape (num_boxes,) containing the IoU values for each pair of ground truth and predicted box.\n",
        "      In case of division by zero (no intersection), the corresponding IoU value will be zero.\n",
        "  \"\"\"\n",
        "\n",
        "  # Split coordinates for efficient calculation\n",
        "  x11, y11, x12, y12 = jnp.split(gt_boxes, 4, axis=-1)\n",
        "  x21, y21, x22, y22 = jnp.split(boxes, 4, axis=-1)\n",
        "\n",
        "  # Calculate overlapping area\n",
        "  x1, y1 = jnp.maximum(x11, jnp.transpose(x21)), jnp.maximum(y11, jnp.transpose(y21))\n",
        "  x2, y2 = jnp.minimum(x12, jnp.transpose(x22)), jnp.minimum(y12, jnp.transpose(y22))\n",
        "  intersection_areas = jnp.maximum(0, (y2 - y1)) * jnp.maximum(0, (x2 - x1))\n",
        "\n",
        "  # Calculate areas of ground truth and predicted boxes\n",
        "  gt_boxes_areas = (x12 - x11) * (y12 - y11)\n",
        "  boxes_areas = (x22 - x21) * (y22 - y21)\n",
        "\n",
        "  # Calculate union areas and handle division by zero\n",
        "  unions = gt_boxes_areas + boxes_areas.reshape(1, -1) - intersection_areas\n",
        "\n",
        "  # Create a maks for intersection area for zero and non-zero values,\n",
        "  # now divide if there are non zero intersection or\n",
        "  # else make the elements zeros without any division.\n",
        "  iou = jnp.where(\n",
        "      jnp.equal(intersection_areas, 0.0),\n",
        "      jnp.zeros_like(intersection_areas),\n",
        "      jnp.true_divide(intersection_areas, unions)\n",
        "  )\n",
        "  return iou\n"
      ],
      "metadata": {
        "id": "keXmqHZ37Ph6"
      },
      "execution_count": 70,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "gt_boxes = jnp.array([\n",
        "    [50, 72, 197, 121],\n",
        "    [39, 63, 203, 112],\n",
        "    [49, 75, 203, 125],\n",
        "    [31, 69, 201, 125],\n",
        "    [50, 72, 197, 121]\n",
        "]) # (Batch_Size, 4)\n",
        "\n",
        "boxes = jnp.array([\n",
        "    [20, 20, 90, 120],\n",
        "    [10, 10, 40, 40],\n",
        "    [42, 78, 186, 126],\n",
        "    [18, 63, 235, 135],\n",
        "    [54, 72, 198, 120]\n",
        "])\n",
        "\n",
        "iou_scores = iou(gt_boxes, boxes)\n",
        "\n",
        "print(f\"Iou Scores:\\n {iou_scores * 100}\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rPlpJ8rtVLm0",
        "outputId": "05399c30-4d78-47a5-eb52-bb9d527055e7"
      },
      "execution_count": 71,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Iou Scores:\n",
            " [[15.63136042  0.         70.7390831  46.10215054 94.66280513]\n",
            " [19.93299832  0.         48.70672503 51.43369176 62.69046582]\n",
            " [14.35239207  0.         78.78380032 49.28315412 79.68519429]\n",
            " [22.27074236  0.         70.03311258 60.93189964 72.60504202]\n",
            " [15.63136042  0.         70.7390831  46.10215054 94.66280513]]\n"
          ]
        }
      ]
    }
  ]
}