feat/intersection-over-union (#15)

* add notebook NiN

* added notebook NiN

* added notebook

* suggest

* add image IoU

* Add IoU on metrics

* Add IoU on metrics

automation/notebooks-table-data.csv

@@ -8,3 +8,4 @@ U-Net, architectures/unet.ipynb,,1505.04597
 DenseNet, architectures/densenet.ipynb,,1608.06993v5
 Mobilenet V1, architectures/mobilenetv1.ipynb,,1704.04861
 Inception V1, architectures/inceptionv1.ipynb,,1409.4842
+Intersection Over Union (IoU),metrics/intersection-over-union.ipynb,,1902.09630

notebooks/metrics/intersection-over-union.ipynb

@@ -0,0 +1,193 @@
+{
+  "nbformat": 4,
+  "nbformat_minor": 0,
+  "metadata": {
+    "colab": {
+      "provenance": []
+    },
+    "kernelspec": {
+      "name": "python3",
+      "display_name": "Python 3"
+    },
+    "language_info": {
+      "name": "python"
+    }
+  },
+  "cells": [
+    {
+      "cell_type": "markdown",
+      "source": [
+        "<image src=\"https://raw.githubusercontent.com/semilleroCV/deep-learning-notes/main/assets/banner-notebook.png\" width=100%>"
+      ],
+      "metadata": {
+        "id": "M1wwX3eCyuPO"
+      }
+    },
+    {
+      "cell_type": "markdown",
+      "source": [
+        "# <font color='#4C5FDA'> **Intersection Over Union (IoU)** </font>"
+      ],
+      "metadata": {
+        "id": "MaBAcPfI0RKH"
+      }
+    },
+    {
+      "cell_type": "markdown",
+      "source": [
+        "The paper **<font color=\"EB9A54\">\"The PASCAL Visual Object Classes (VOC) Challenge\"</font>** introduces Intersection over Union (IoU) as a critical metric for evaluating object detection models. IoU measures the overlap between predicted and ground truth bounding boxes, providing a robust standard for assessing model accuracy. Building on this foundation, the paper **<font color=\"EB9A54\">\"Generalized Intersection over Union: A Metric and A Loss for Bounding Box Regression\"</font>** extends the concept by introducing Generalized IoU (GIoU), which improves model training and performance by offering a more comprehensive evaluation of bounding box regression."
+      ],
+      "metadata": {
+        "id": "iEV5ns6T0Ynt"
+      }
+    },
+    {
+      "cell_type": "markdown",
+      "source": [
+        "$$IOU(\\mathcal{A},\\mathcal{B}) = \\frac{\\left|\\mathcal{A} \\cap \\mathcal{B}\\right|}{\\left| \\mathcal{A} \\cup \\mathcal{B}\\right|}.$$"
+      ],
+      "metadata": {
+        "id": "18LUav2hcNlG"
+      }
+    },
+    {
+      "cell_type": "markdown",
+      "source": [
+        "The mAP(mean average precision) metric in object detection is evaluated based on IoU (Intersection Over Union)."
+      ],
+      "metadata": {
+        "id": "bDjg6i60bdep"
+      }
+    },
+    {
+      "cell_type": "markdown",
+      "source": [
+        "<image src=\"https://raw.githubusercontent.com/semilleroCV/deep-learning-notes/feat/intersection-over-union/assets/feature-image-iou.jpg\" width=100%>"
+      ],
+      "metadata": {
+        "id": "Upy9-GAbcT-v"
+      }
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 22,
+      "metadata": {
+        "id": "xYEl0lf0v2vl"
+      },
+      "outputs": [],
+      "source": [
+        "import torch\n",
+        "import torch.nn as nn\n",
+        "import torch.nn.functional as F"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "print(torch.__version__)"
+      ],
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "6_uXqqwdedqU",
+        "outputId": "158eded7-83c3-4bd0-ab19-97b1c97428fb"
+      },
+      "execution_count": 23,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "2.3.0+cu121\n"
+          ]
+        }
+      ]
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "import torch\n",
+        "import torch.nn as nn\n",
+        "\n",
+        "class IoU(nn.Module):\n",
+        "    def __init__(self):\n",
+        "        super(IoU, self).__init__()\n",
+        "\n",
+        "    def forward(self, box1, box2):\n",
+        "        # Calculate intersection\n",
+        "        x1 = torch.max(box1[0], box2[0])\n",
+        "        y1 = torch.max(box1[1], box2[1])\n",
+        "        x2 = torch.min(box1[2], box2[2])\n",
+        "        y2 = torch.min(box1[3], box2[3])\n",
+        "\n",
+        "        intersection = (x2 - x1).clamp(0) * (y2 - y1).clamp(0)\n",
+        "\n",
+        "        # Calculate areas in the box\n",
+        "        box1_area = (box1[2] - box1[0]) * (box1[3] - box1[1])\n",
+        "        box2_area = (box2[2] - box2[0]) * (box2[3] - box2[1])\n",
+        "\n",
+        "        # Calculate union\n",
+        "        union = box1_area + box2_area - intersection\n",
+        "\n",
+        "        # Calculate IoU\n",
+        "        iou = intersection / union\n",
+        "        return iou\n"
+      ],
+      "metadata": {
+        "id": "FWOemCQ8g0P4"
+      },
+      "execution_count": 33,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "# Example usage 1\n",
+        "box1 = torch.tensor([10, 20, 25, 35], dtype=torch.float)\n",
+        "box2 = torch.tensor([10, 20, 30, 40], dtype=torch.float)\n",
+        "\n",
+        "iou_calculator = IoU()\n",
+        "iou = iou_calculator(box1, box2)\n",
+        "print(f'IoU: {iou.item()}')"
+      ],
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "NCCqOcI3g9l-",
+        "outputId": "5846553d-50b4-45c9-8c0c-51910f92745c"
+      },
+      "execution_count": 34,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "IoU: 0.5625\n"
+          ]
+        }
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "source": [
+        "---\n",
+        "# <font color='#4C5FDA'> **References** </font>\n",
+        "\n",
+        "The PASCAL Visual Object Classes (VOC) Challenge\n",
+        "\n",
+        "https://link.springer.com/article/10.1007/s11263-009-0275-4\n",
+        "\n",
+        "Generalized Intersection over Union: A Metric and A Loss for Bounding Box Regression\n",
+        "\n",
+        "https://arxiv.org/pdf/1902.09630\n",
+        "\n",
+        "---"
+      ],
+      "metadata": {
+        "id": "g81tr2oajhtQ"
+      }
+    }
+  ]
+}