pytorch initial commit

keypoint_classification_EN_pytorch.ipynb

@@ -0,0 +1,383 @@
+{
+  "cells": [
+    {
+      "cell_type": "code",
+      "execution_count": 1,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "ypqky9tc9hE1",
+        "outputId": "5db082bb-30e3-4110-bf63-a1ee777ecd46"
+      },
+      "outputs": [
+        {
+          "name": "stdout",
+          "output_type": "stream",
+          "text": [
+            "Collecting torch\n",
+            "  Using cached torch-2.9.1-cp312-cp312-win_amd64.whl.metadata (30 kB)\n",
+            "Collecting torchvision\n",
+            "  Using cached torchvision-0.24.1-cp312-cp312-win_amd64.whl.metadata (5.9 kB)\n",
+            "Collecting filelock (from torch)\n",
+            "  Using cached filelock-3.20.0-py3-none-any.whl.metadata (2.1 kB)\n",
+            "Requirement already satisfied: typing-extensions>=4.10.0 in c:\\users\\rfd\\appdata\\local\\programs\\python\\python312\\lib\\site-packages (from torch) (4.15.0)\n",
+            "Requirement already satisfied: sympy>=1.13.3 in c:\\users\\rfd\\appdata\\local\\programs\\python\\python312\\lib\\site-packages (from torch) (1.14.0)\n",
+            "Collecting networkx>=2.5.1 (from torch)\n",
+            "  Using cached networkx-3.5-py3-none-any.whl.metadata (6.3 kB)\n",
+            "Collecting jinja2 (from torch)\n",
+            "  Using cached jinja2-3.1.6-py3-none-any.whl.metadata (2.9 kB)\n",
+            "Collecting fsspec>=0.8.5 (from torch)\n",
+            "  Using cached fsspec-2025.10.0-py3-none-any.whl.metadata (10 kB)\n",
+            "Collecting setuptools (from torch)\n",
+            "  Using cached setuptools-80.9.0-py3-none-any.whl.metadata (6.6 kB)\n",
+            "Collecting numpy (from torchvision)\n",
+            "  Using cached numpy-2.3.4-cp312-cp312-win_amd64.whl.metadata (60 kB)\n",
+            "Collecting pillow!=8.3.*,>=5.3.0 (from torchvision)\n",
+            "  Using cached pillow-12.0.0-cp312-cp312-win_amd64.whl.metadata (9.0 kB)\n",
+            "Requirement already satisfied: mpmath<1.4,>=1.1.0 in c:\\users\\rfd\\appdata\\local\\programs\\python\\python312\\lib\\site-packages (from sympy>=1.13.3->torch) (1.3.0)\n",
+            "Collecting MarkupSafe>=2.0 (from jinja2->torch)\n",
+            "  Using cached markupsafe-3.0.3-cp312-cp312-win_amd64.whl.metadata (2.8 kB)\n",
+            "Using cached torch-2.9.1-cp312-cp312-win_amd64.whl (110.9 MB)\n",
+            "Using cached torchvision-0.24.1-cp312-cp312-win_amd64.whl (4.3 MB)\n",
+            "Using cached fsspec-2025.10.0-py3-none-any.whl (200 kB)\n",
+            "Using cached networkx-3.5-py3-none-any.whl (2.0 MB)\n",
+            "Using cached pillow-12.0.0-cp312-cp312-win_amd64.whl (7.0 MB)\n",
+            "Using cached filelock-3.20.0-py3-none-any.whl (16 kB)\n",
+            "Using cached jinja2-3.1.6-py3-none-any.whl (134 kB)\n",
+            "Using cached numpy-2.3.4-cp312-cp312-win_amd64.whl (12.8 MB)\n",
+            "Using cached setuptools-80.9.0-py3-none-any.whl (1.2 MB)\n",
+            "Using cached markupsafe-3.0.3-cp312-cp312-win_amd64.whl (15 kB)\n",
+            "Installing collected packages: setuptools, pillow, numpy, networkx, MarkupSafe, fsspec, filelock, jinja2, torch, torchvision\n",
+            "Successfully installed MarkupSafe-3.0.3 filelock-3.20.0 fsspec-2025.10.0 jinja2-3.1.6 networkx-3.5 numpy-2.3.4 pillow-12.0.0 setuptools-80.9.0 torch-2.9.1 torchvision-0.24.1\n"
+          ]
+        },
+        {
+          "name": "stderr",
+          "output_type": "stream",
+          "text": [
+            "\n",
+            "[notice] A new release of pip is available: 24.2 -> 25.3\n",
+            "[notice] To update, run: python.exe -m pip install --upgrade pip\n"
+          ]
+        }
+      ],
+      "source": [
+        "!pip install torch torchvision\n",
+        "import torch\n",
+        "import torch.nn as nn\n",
+        "import torch.nn.functional as F\n",
+        "\n",
+        "class KeypointClassifier(nn.Module):\n",
+        "    def __init__(self, num_classes=12):\n",
+        "        super().__init__()\n",
+        "        self.dropout1 = nn.Dropout(0.2)\n",
+        "        self.fc1 = nn.Linear(42, 20)\n",
+        "        self.dropout2 = nn.Dropout(0.4)\n",
+        "        self.fc2 = nn.Linear(20, 10)\n",
+        "        self.fc3 = nn.Linear(10, num_classes)\n",
+        "\n",
+        "    def forward(self, x):\n",
+        "        x = self.dropout1(x)\n",
+        "        x = F.relu(self.fc1(x))\n",
+        "        x = self.dropout2(x)\n",
+        "        x = F.relu(self.fc2(x))\n",
+        "        x = self.fc3(x)   # NO softmax here (PyTorch loss expects raw logits)\n",
+        "        return x\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "MbMjOflQ9hE1"
+      },
+      "outputs": [
+        {
+          "name": "stdout",
+          "output_type": "stream",
+          "text": [
+            "Collecting sklearn\n",
+            "  Downloading sklearn-0.0.post12.tar.gz (2.6 kB)\n",
+            "  Installing build dependencies: started\n",
+            "  Installing build dependencies: finished with status 'done'\n",
+            "  Getting requirements to build wheel: started\n",
+            "  Getting requirements to build wheel: finished with status 'error'\n"
+          ]
+        },
+        {
+          "name": "stderr",
+          "output_type": "stream",
+          "text": [
+            "  error: subprocess-exited-with-error\n",
+            "  \n",
+            "  × Getting requirements to build wheel did not run successfully.\n",
+            "  │ exit code: 1\n",
+            "  ╰─> [15 lines of output]\n",
+            "      The 'sklearn' PyPI package is deprecated, use 'scikit-learn'\n",
+            "      rather than 'sklearn' for pip commands.\n",
+            "      \n",
+            "      Here is how to fix this error in the main use cases:\n",
+            "      - use 'pip install scikit-learn' rather than 'pip install sklearn'\n",
+            "      - replace 'sklearn' by 'scikit-learn' in your pip requirements files\n",
+            "        (requirements.txt, setup.py, setup.cfg, Pipfile, etc ...)\n",
+            "      - if the 'sklearn' package is used by one of your dependencies,\n",
+            "        it would be great if you take some time to track which package uses\n",
+            "        'sklearn' instead of 'scikit-learn' and report it to their issue tracker\n",
+            "      - as a last resort, set the environment variable\n",
+            "        SKLEARN_ALLOW_DEPRECATED_SKLEARN_PACKAGE_INSTALL=True to avoid this error\n",
+            "      \n",
+            "      More information is available at\n",
+            "      https://github.com/scikit-learn/sklearn-pypi-package\n",
+            "      [end of output]\n",
+            "  \n",
+            "  note: This error originates from a subprocess, and is likely not a problem with pip.\n",
+            "\n",
+            "[notice] A new release of pip is available: 24.2 -> 25.3\n",
+            "[notice] To update, run: python.exe -m pip install --upgrade pip\n",
+            "error: subprocess-exited-with-error\n",
+            "\n",
+            "× Getting requirements to build wheel did not run successfully.\n",
+            "│ exit code: 1\n",
+            "╰─> See above for output.\n",
+            "\n",
+            "note: This error originates from a subprocess, and is likely not a problem with pip.\n"
+          ]
+        },
+        {
+          "ename": "ModuleNotFoundError",
+          "evalue": "No module named 'sklearn'",
+          "output_type": "error",
+          "traceback": [
+            "\u001b[31m---------------------------------------------------------------------------\u001b[39m",
+            "\u001b[31mModuleNotFoundError\u001b[39m                       Traceback (most recent call last)",
+            "\u001b[36mCell\u001b[39m\u001b[36m \u001b[39m\u001b[32mIn[3]\u001b[39m\u001b[32m, line 4\u001b[39m\n\u001b[32m      1\u001b[39m get_ipython().system(\u001b[33m'\u001b[39m\u001b[33mpip install sklearn\u001b[39m\u001b[33m'\u001b[39m)\n\u001b[32m      3\u001b[39m \u001b[38;5;28;01mimport\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[34;01mnumpy\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[38;5;28;01mas\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[34;01mnp\u001b[39;00m\n\u001b[32m----> \u001b[39m\u001b[32m4\u001b[39m \u001b[38;5;28;01mfrom\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[34;01msklearn\u001b[39;00m\u001b[34;01m.\u001b[39;00m\u001b[34;01mmodel_selection\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[38;5;28;01mimport\u001b[39;00m train_test_split\n\u001b[32m      5\u001b[39m \u001b[38;5;28;01mimport\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[34;01mtorch\u001b[39;00m\n\u001b[32m      6\u001b[39m \u001b[38;5;28;01mfrom\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[34;01mtorch\u001b[39;00m\u001b[34;01m.\u001b[39;00m\u001b[34;01mutils\u001b[39;00m\u001b[34;01m.\u001b[39;00m\u001b[34;01mdata\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[38;5;28;01mimport\u001b[39;00m TensorDataset, DataLoader\n",
+            "\u001b[31mModuleNotFoundError\u001b[39m: No module named 'sklearn'"
+          ]
+        }
+      ],
+      "source": [
+        "!pip install scikit-learn\n",
+        "\n",
+        "import numpy as np\n",
+        "from sklearn.model_selection import train_test_split\n",
+        "import torch\n",
+        "from torch.utils.data import TensorDataset, DataLoader\n",
+        "\n",
+        "dataset = 'model/keypoint_classifier/keypoint.csv'\n",
+        "\n",
+        "X = np.loadtxt(dataset, delimiter=',', dtype='float32', usecols=list(range(1, 43)))\n",
+        "y = np.loadtxt(dataset, delimiter=',', dtype='int64', usecols=(0))\n",
+        "\n",
+        "X_train, X_test, y_train, y_test = train_test_split(\n",
+        "    X, y, train_size=0.75, random_state=42\n",
+        ")\n",
+        "\n",
+        "train_ds = TensorDataset(\n",
+        "    torch.tensor(X_train, dtype=torch.float32),\n",
+        "    torch.tensor(y_train, dtype=torch.long)\n",
+        ")\n",
+        "test_ds = TensorDataset(\n",
+        "    torch.tensor(X_test, dtype=torch.float32),\n",
+        "    torch.tensor(y_test, dtype=torch.long)\n",
+        ")\n",
+        "\n",
+        "train_dl = DataLoader(train_ds, batch_size=128, shuffle=True)\n",
+        "test_dl = DataLoader(test_ds, batch_size=128)\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "c3Dac0M_9hE2"
+      },
+      "outputs": [],
+      "source": [
+        "model = KeypointClassifier(num_classes=12)\n",
+        "optimizer = torch.optim.Adam(model.parameters(), lr=1e-3)\n",
+        "criterion = nn.CrossEntropyLoss()\n",
+        "\n",
+        "EPOCHS = 100\n",
+        "\n",
+        "for epoch in range(EPOCHS):\n",
+        "    model.train()\n",
+        "    total_loss = 0\n",
+        "\n",
+        "    for xb, yb in train_dl:\n",
+        "        optimizer.zero_grad()\n",
+        "        logits = model(xb)\n",
+        "        loss = criterion(logits, yb)\n",
+        "        loss.backward()\n",
+        "        optimizer.step()\n",
+        "        total_loss += loss.item()\n",
+        "\n",
+        "    print(f\"Epoch {epoch+1}/{EPOCHS} - Loss: {total_loss:.4f}\")\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "WirBl-JE9hE3",
+        "outputId": "71b30ca2-8294-4d9d-8aa2-800d90d399de",
+        "scrolled": true
+      },
+      "outputs": [],
+      "source": [
+        "model.eval()\n",
+        "correct = 0\n",
+        "total = 0\n",
+        "\n",
+        "with torch.no_grad():\n",
+        "    for xb, yb in test_dl:\n",
+        "        preds = model(xb).argmax(dim=1)\n",
+        "        correct += (preds == yb).sum().item()\n",
+        "        total += yb.size(0)\n",
+        "\n",
+        "print(\"Accuracy:\", correct / total)\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "pxvb2Y299hE3",
+        "outputId": "59eb3185-2e37-4b9e-bc9d-ab1b8ac29b7f"
+      },
+      "outputs": [],
+      "source": [
+        "torch.save(model.state_dict(), \"keypoint_classifier_pytorch.pth\")\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "RBkmDeUW9hE4"
+      },
+      "outputs": [],
+      "source": [
+        "model_quant = torch.quantization.quantize_dynamic(\n",
+        "    model, \n",
+        "    {nn.Linear},     # Quantize only linear layers\n",
+        "    dtype=torch.qint8\n",
+        ")\n",
+        "\n",
+        "torch.save(model_quant.state_dict(), \"keypoint_classifier_quantized.pth\")\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "tFz9Tb0I9hE4",
+        "outputId": "1c3b3528-54ae-4ee2-ab04-77429211cbef"
+      },
+      "outputs": [],
+      "source": []
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/",
+          "height": 582
+        },
+        "id": "AP1V6SCk9hE5",
+        "outputId": "08e41a80-7a4a-4619-8125-ecc371368d19"
+      },
+      "outputs": [],
+      "source": []
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "ODjnYyld9hE6"
+      },
+      "outputs": [],
+      "source": []
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "zRfuK8Y59hE6",
+        "outputId": "a4ca585c-b5d5-4244-8291-8674063209bb"
+      },
+      "outputs": [],
+      "source": []
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "s4FoAnuc9hE7",
+        "outputId": "91f18257-8d8b-4ef3-c558-e9b5f94fabbf",
+        "scrolled": true
+      },
+      "outputs": [],
+      "source": [
+        "\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "vONjp19J9hE8",
+        "outputId": "77205e24-fd00-42c4-f7b6-e06e527c2cba"
+      },
+      "outputs": [],
+      "source": []
+    }
+  ],
+  "metadata": {
+    "accelerator": "GPU",
+    "colab": {
+      "collapsed_sections": [],
+      "name": "keypoint_classification_EN.ipynb",
+      "provenance": [],
+      "toc_visible": true
+    },
+    "kernelspec": {
+      "display_name": "Python 3",
+      "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": "3.12.7"
+    }
+  },
+  "nbformat": 4,
+  "nbformat_minor": 0
+}

keypoint_classifier_label.csv

@@ -0,0 +1,12 @@
+Open
+Close
+Pointer
+Pinch
+Thumbs Up
+Thumbs Down
+Thumbs Sideways
+Pinch Pinky
+L
+Click Up
+Click Down
+Yolo

keypoint_classifier_pytorch.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2e2939ce6244459e743d8d098585051961be0c01761be81bd606fcb6731086b
+size 8005

keypoint_classifier_pytorch.py

@@ -0,0 +1,42 @@
+import torch
+import torch.nn as nn
+
+# -------------------------------
+# PyTorch model definition
+# -------------------------------
+class KeyPointClassifierModel(nn.Module):
+    def __init__(self):
+        super().__init__()
+        self.fc1 = nn.Linear(42, 20)
+        self.relu1 = nn.ReLU()
+        self.fc2 = nn.Linear(20, 10)
+        self.relu2 = nn.ReLU()
+        self.fc3 = nn.Linear(10, 12)  # match checkpoint output classes
+
+    def forward(self, x):
+        x = self.fc1(x)
+        x = self.relu1(x)
+        x = self.fc2(x)
+        x = self.relu2(x)
+        x = self.fc3(x)
+        return x
+
+# -------------------------------
+# Wrapper class for easy usage
+# -------------------------------
+class KeyPointClassifier:
+    def __init__(self, model_path="keypoint_classifier_pytorch.pth", device='cpu'):
+        self.device = device
+        self.model = KeyPointClassifierModel()
+        # Load the checkpoint
+        self.model.load_state_dict(torch.load(model_path, map_location=self.device))
+        self.model.to(self.device)
+        self.model.eval()
+
+    def __call__(self, landmark_list):
+        with torch.no_grad():
+            x = torch.tensor([landmark_list], dtype=torch.float32).to(self.device)
+            output = self.model(x)
+            prob = torch.softmax(output, dim=1)
+            conf, pred = torch.max(prob, dim=1)
+            return pred.item(), conf.item()

keypoint_classifier_quantized.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d06732f994e873e78791a9785ef120deb2e8ab30fdc066741ea4900bae92443
+size 6731