Upload 7 files

README.md

@@ -1,12 +1,13 @@
 ---
-title: Ham1000
-emoji: 📊
-colorFrom: blue
-colorTo: blue
+title: Bodypartxr
+emoji: 🏆
+colorFrom: red
+colorTo: pink
 sdk: gradio
-sdk_version: 3.50.2
+sdk_version: 3.47.1
 app_file: app.py
 pinned: false
+license: unknown
 ---
 
 Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -0,0 +1,265 @@
+import gradio as gr
+import torch
+from torchvision.transforms import transforms
+import numpy as np
+from typing import Optional
+import torch.nn as nn
+import os
+from utils import page_utils
+
+class BasicBlock(nn.Module):
+    """ResNet Basic Block.
+
+    Parameters
+    ----------
+    in_channels : int
+        Number of input channels
+    out_channels : int
+        Number of output channels
+    stride : int, optional
+        Convolution stride size, by default 1
+    identity_downsample : Optional[torch.nn.Module], optional
+        Downsampling layer, by default None
+    """
+
+    def __init__(self,
+                in_channels: int,
+                out_channels: int,
+                stride: int = 1,
+                identity_downsample: Optional[torch.nn.Module] = None):
+        super(BasicBlock, self).__init__()
+        self.conv1 = nn.Conv2d(in_channels,
+                              out_channels,
+                              kernel_size = 3,
+                              stride = stride,
+                              padding = 1)
+        self.bn1 = nn.BatchNorm2d(out_channels)
+        self.relu = nn.ReLU()
+        self.conv2 = nn.Conv2d(out_channels,
+                              out_channels,
+                              kernel_size = 3,
+                              stride = 1,
+                              padding = 1)
+        self.bn2 = nn.BatchNorm2d(out_channels)
+        self.identity_downsample = identity_downsample
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        """Apply forward computation."""
+        identity = x
+        x = self.conv1(x)
+        x = self.bn1(x)
+        x = self.relu(x)
+        x = self.conv2(x)
+        x = self.bn2(x)
+
+        # Apply an operation to the identity output.
+        # Useful to reduce the layer size and match from conv2 output
+        if self.identity_downsample is not None:
+            identity = self.identity_downsample(identity)
+        x += identity
+        x = self.relu(x)
+        return x
+
+class ResNet18(nn.Module):
+    """Construct ResNet-18 Model.
+
+    Parameters
+    ----------
+    input_channels : int
+        Number of input channels
+    num_classes : int
+        Number of class outputs
+    """
+
+    def __init__(self, input_channels, num_classes):
+
+        super(ResNet18, self).__init__()
+        self.conv1 = nn.Conv2d(input_channels,
+                               64, kernel_size = 7,
+                              stride = 2, padding=3)
+        self.bn1 = nn.BatchNorm2d(64)
+        self.relu = nn.ReLU()
+        self.maxpool = nn.MaxPool2d(kernel_size = 3,
+                                   stride = 2,
+                                   padding = 1)
+
+        self.layer1 = self._make_layer(64, 64, stride = 1)
+        self.layer2 = self._make_layer(64, 128, stride = 2)
+        self.layer3 = self._make_layer(128, 256, stride = 2)
+        self.layer4 = self._make_layer(256, 512, stride = 2)
+
+        # Last layers
+        self.avgpool = nn.AdaptiveAvgPool2d((1, 1))
+        self.fc = nn.Linear(512, num_classes)
+
+    def identity_downsample(self, in_channels: int, out_channels: int) -> nn.Module:
+        """Downsampling block to reduce the feature sizes."""
+        return nn.Sequential(
+             nn.Conv2d(in_channels,
+                       out_channels,
+                       kernel_size = 3,
+                       stride = 2,
+                       padding = 1),
+            nn.BatchNorm2d(out_channels)
+        )
+
+    def _make_layer(self, in_channels: int, out_channels: int, stride: int) -> nn.Module:
+        """Create sequential basic block."""
+        identity_downsample = None
+
+        # Add downsampling function
+        if stride != 1:
+            identity_downsample = self.identity_downsample(in_channels, out_channels)
+
+        return nn.Sequential(
+                    BasicBlock(in_channels, out_channels, identity_downsample=identity_downsample, stride=stride),
+                    BasicBlock(out_channels, out_channels)
+                    )
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        x = self.conv1(x)
+        x = self.bn1(x)
+        x = self.relu(x)
+        x = self.maxpool(x)
+
+        x = self.layer1(x)
+        x = self.layer2(x)
+        x = self.layer3(x)
+        x = self.layer4(x)
+
+        x = self.avgpool(x)
+        x = x.view(x.shape[0], -1)
+        x = self.fc(x)
+        return x
+
+model = ResNet18(1, 7)
+
+checkpoint = torch.load('ham1.ckpt', map_location=torch.device('cpu'))
+
+# The state dict will contains net.layer_name
+# Our model doesn't contains `net.` so we have to rename it
+state_dict = checkpoint['state_dict']
+for key in list(state_dict.keys()):
+    if 'net.' in key:
+        state_dict[key.replace('net.', '')] = state_dict[key]
+        del state_dict[key]
+
+model.load_state_dict(state_dict)
+model.eval()
+
+class_names = ['akk', 'bcc', 'bkl', 'df', 'mel','nv','vasc']
+class_names.sort()
+
+examples_dir = "sample"
+
+transformation_pipeline = transforms.Compose([
+    transforms.ToPILImage(),
+    transforms.Grayscale(num_output_channels=1),
+    transforms.CenterCrop((224, 224)),
+    transforms.ToTensor(),
+    transforms.Normalize(mean=[0.485], std=[0.229])
+])
+
+
+def preprocess_image(image: np.ndarray):
+    """Preprocess the input image.
+
+    Note that the input image is in RGB mode.
+
+    Parameters
+    ----------
+    image: np.ndarray
+        Input image from callback.
+    """
+
+    image = transformation_pipeline(image)
+    image = torch.unsqueeze(image, 0)
+
+    return image
+
+
+def image_classifier(inp):
+    """Image Classifier Function.
+
+    Parameters
+    ----------
+    inp: Optional[np.ndarray] = None
+        Input image from callback
+
+    Returns
+    -------
+    Dict
+        A dictionary class names and its probability
+    """
+
+    # If input not valid, return dummy data or raise error
+    if inp is None:
+        return {'cat': 0.3, 'dog': 0.7}
+
+    # preprocess
+    image = preprocess_image(inp)
+    image = image.to(dtype=torch.float32)
+
+    # inference
+    result = model(image)
+
+    # postprocess
+    result = torch.nn.functional.softmax(result, dim=1) # apply softmax
+    result = result[0].detach().numpy().tolist() # take the first batch
+    labeled_result = {name:score for name, score in zip(class_names, result)}
+
+    return labeled_result
+
+# gradio code block for input and output
+with gr.Blocks() as app:
+    gr.Markdown("# Skin Cancer Classification")
+
+with open('index.html', encoding="utf-8") as f:
+    description = f.read()
+
+# gradio code block for input and output
+with gr.Blocks(theme=gr.themes.Default(primary_hue=page_utils.KALBE_THEME_COLOR, secondary_hue=page_utils.KALBE_THEME_COLOR).set(
+        button_primary_background_fill="*primary_600",
+        button_primary_background_fill_hover="*primary_500",
+        button_primary_text_color="white",
+    )) as app:
+    with gr.Column():
+        gr.HTML(description)
+
+    with gr.Row():
+        with gr.Column():
+            inp_img = gr.Image()
+            with gr.Row():
+                clear_btn = gr.Button(value="Clear")
+                process_btn = gr.Button(value="Process", variant="primary")
+        with gr.Column():
+            out_txt = gr.Label(label="Probabilities", num_top_classes=3)
+
+    process_btn.click(image_classifier, inputs=inp_img, outputs=out_txt)
+    clear_btn.click(lambda:(
+        gr.update(value=None),
+        gr.update(value=None)
+        ),
+        inputs=None,
+        outputs=[inp_img, out_txt])
+
+    gr.Markdown("## Image Examples")
+    gr.Examples(
+        examples=[os.path.join(examples_dir, "ISIC_0000108_downsampled.jpeg"),
+                  os.path.join(examples_dir, "ISIC_0000142_downsampled.jpeg"),
+                  os.path.join(examples_dir, "ISIC_0012792_downsampled.jpeg"),
+                  os.path.join(examples_dir, "ISIC_0024452.jpeg"),
+                  os.path.join(examples_dir, "ISIC_0025957.jpeg"),
+                  os.path.join(examples_dir, "ISIC_0026876.jpeg"),
+                  os.path.join(examples_dir, "ISIC_0027385.jpeg"),
+                  os.path.join(examples_dir, "ISIC_0030956.jpeg"),
+                 ],
+        inputs=inp_img,
+        outputs=out_txt,
+        fn=image_classifier,
+        cache_examples=False,
+    )
+    gr.Markdown(line_breaks=True, value='Author: Jason Adrian ([email protected]) <div class="row"><a href="https://github.com/jasonadriann?tab=repositories"><img alt="GitHub" src="https://img.shields.io/badge/Jason%20Adrian-000000?logo=github"> </div>')
+
+# demo = gr.Interface(fn=image_classifier, inputs="image", outputs="label")
+app.launch(share=True)

ham1.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b07ac05dfb7cb1b0f0d57ad5baf923acd0d4da5352588ae492f4faa970e2833
+size 150928119

index.html

@@ -0,0 +1,50 @@
+<!DOCTYPE html>
+<html>
+	<head>
+		<link rel="stylesheet" href="file/style.css" />
+		<link rel="preconnect" href="https://fonts.googleapis.com" />
+		<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin />
+		<link href="https://fonts.googleapis.com/css2?family=Source+Sans+Pro:wght@400;600;700&display=swap" rel="stylesheet" />
+		<title><strong>Body Part Classification</strong></title>
+	</head>
+	<body>
+		<div class="container">
+			<h1 class="title"><strong> Body Part Classification</strong></h1>
+			<h2 class="subtitle"><strong>Kalbe Digital Lab</strong></h2>
+			<section class="overview">
+				<div class="grid-container">
+					<h3 class="overview-heading"><span class="vl">Overview</span></h3>
+					<p class="overview-content">
+						The Body Part Classification program serves the critical purpose of categorizing body parts from DICOM x-ray scans into five distinct classes: abdominal, adult chest, pediatric chest, spine, and others. This program trained using ResNet18 model.
+					</p>
+				</div>
+				<div class="grid-container">
+					<h3 class="overview-heading"><span class="vl">Dataset</span></h3>
+					<div>
+						<p class="overview-content">
+						The program has been meticulously trained on a robust and diverse dataset, specifically <a href="https://vindr.ai/datasets/bodypartxr" target="_blank">VinDrBodyPartXR Dataset.</a>. 
+						<br/>
+						This dataset is introduced by Vingroup of Big Data Institute which include 16,093 x-ray images that are collected and manually annotated. It is a highly valuable resource that has been instrumental in the training of our model.
+						</p>
+						<ul>
+							<li>Objective: Body Part Identification</li>
+							<li>Task: Classification</li>
+							<li>Modality: Grayscale Images</li>
+						</ul>
+					</div>
+				</div>
+				<div class="grid-container">
+					<h3 class="overview-heading"><span class="vl">Model Architecture</span></h3>
+					<div>
+						<p class="overview-content">
+							The model architecture of ResNet18 to train x-ray images for classifying body part.
+						</p>
+						<img class="content-image" src="file/figures/ResNet-18.png" alt="model-architecture" width="425" height="115" style="vertical-align:middle" />
+					</div>
+				</div>
+			</section>
+			<h3 class="overview-heading"><span class="vl">Demo</span></h3>
+			<p class="overview-content">Please select or upload a body part x-ray scan image to see the capabilities of body part classification with this model</p>
+		</div>
+	</body>
+</html>

requirements.txt

@@ -0,0 +1,63 @@
+aiofiles==23.2.1
+altair==5.1.2
+annotated-types==0.6.0
+anyio==3.7.1
+attrs==23.1.0
+certifi==2023.7.22
+charset-normalizer==3.3.0
+click==8.1.7
+colorama==0.4.6
+contourpy==1.1.1
+cycler==0.12.1
+exceptiongroup==1.1.3
+fastapi==0.103.2
+ffmpy==0.3.1
+filelock==3.12.4
+fonttools==4.43.1
+fsspec==2023.9.2
+gradio==3.47.1
+gradio_client==0.6.0
+h11==0.14.0
+httpcore==0.18.0
+httpx==0.25.0
+huggingface-hub==0.17.3
+idna==3.4
+importlib-resources==6.1.0
+Jinja2==3.1.2
+jsonschema==4.19.1
+jsonschema-specifications==2023.7.1
+kiwisolver==1.4.5
+MarkupSafe==2.1.3
+matplotlib==3.8.0
+mpmath==1.3.0
+networkx==3.1
+numpy==1.26.0
+orjson==3.9.7
+packaging==23.2
+pandas==2.1.1
+Pillow==10.0.1
+pydantic==2.4.2
+pydantic_core==2.10.1
+pydub==0.25.1
+pyparsing==3.1.1
+python-dateutil==2.8.2
+python-multipart==0.0.6
+pytz==2023.3.post1
+PyYAML==6.0.1
+referencing==0.30.2
+requests==2.31.0
+rpds-py==0.10.4
+semantic-version==2.10.0
+six==1.16.0
+sniffio==1.3.0
+starlette==0.27.0
+sympy==1.12
+toolz==0.12.0
+torch==2.1.0
+torchvision==0.16.0
+tqdm==4.66.1
+typing_extensions==4.8.0
+tzdata==2023.3
+urllib3==2.0.6
+uvicorn==0.23.2
+websockets==11.0.3

resnet18.py

@@ -0,0 +1,129 @@
+from typing import Optional
+
+import torch.nn as nn
+import torch
+
+class BasicBlock(nn.Module):
+    """ResNet Basic Block.
+
+    Parameters
+    ----------
+    in_channels : int
+        Number of input channels
+    out_channels : int
+        Number of output channels
+    stride : int, optional
+        Convolution stride size, by default 1
+    identity_downsample : Optional[torch.nn.Module], optional
+        Downsampling layer, by default None
+    """
+
+    def __init__(self,
+                in_channels: int,
+                out_channels: int,
+                stride: int = 1,
+                identity_downsample: Optional[torch.nn.Module] = None):
+        super(BasicBlock, self).__init__()
+        self.conv1 = nn.Conv2d(in_channels,
+                              out_channels,
+                              kernel_size = 3,
+                              stride = stride,
+                              padding = 1)
+        self.bn1 = nn.BatchNorm2d(out_channels)
+        self.relu = nn.ReLU()
+        self.conv2 = nn.Conv2d(out_channels,
+                              out_channels,
+                              kernel_size = 3,
+                              stride = 1,
+                              padding = 1)
+        self.bn2 = nn.BatchNorm2d(out_channels)
+        self.identity_downsample = identity_downsample
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        """Apply forward computation."""
+        identity = x
+        x = self.conv1(x)
+        x = self.bn1(x)
+        x = self.relu(x)
+        x = self.conv2(x)
+        x = self.bn2(x)
+
+        # Apply an operation to the identity output.
+        # Useful to reduce the layer size and match from conv2 output
+        if self.identity_downsample is not None:
+            identity = self.identity_downsample(identity)
+        x += identity
+        x = self.relu(x)
+        return x
+
+class ResNet18(nn.Module):
+    """Construct ResNet-18 Model.
+
+    Parameters
+    ----------
+    input_channels : int
+        Number of input channels
+    num_classes : int
+        Number of class outputs
+    """
+
+    def __init__(self, input_channels, num_classes):
+
+        super(ResNet18, self).__init__()
+        self.conv1 = nn.Conv2d(input_channels,
+                               64, kernel_size = 7,
+                              stride = 2, padding=3)
+        self.bn1 = nn.BatchNorm2d(64)
+        self.relu = nn.ReLU()
+        self.maxpool = nn.MaxPool2d(kernel_size = 3,
+                                   stride = 2,
+                                   padding = 1)
+
+        self.layer1 = self._make_layer(64, 64, stride = 1)
+        self.layer2 = self._make_layer(64, 128, stride = 2)
+        self.layer3 = self._make_layer(128, 256, stride = 2)
+        self.layer4 = self._make_layer(256, 512, stride = 2)
+
+        # Last layers
+        self.avgpool = nn.AdaptiveAvgPool2d((1, 1))
+        self.fc = nn.Linear(512, num_classes)
+
+    def identity_downsample(self, in_channels: int, out_channels: int) -> nn.Module:
+        """Downsampling block to reduce the feature sizes."""
+        return nn.Sequential(
+             nn.Conv2d(in_channels,
+                       out_channels,
+                       kernel_size = 3,
+                       stride = 2,
+                       padding = 1),
+            nn.BatchNorm2d(out_channels)
+        )
+
+    def _make_layer(self, in_channels: int, out_channels: int, stride: int) -> nn.Module:
+        """Create sequential basic block."""
+        identity_downsample = None
+
+        # Add downsampling function
+        if stride != 1:
+            identity_downsample = self.identity_downsample(in_channels, out_channels)
+
+        return nn.Sequential(
+                    BasicBlock(in_channels, out_channels, identity_downsample=identity_downsample, stride=stride),
+                    BasicBlock(out_channels, out_channels)
+                    )
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        x = self.conv1(x)
+        x = self.bn1(x)
+        x = self.relu(x)
+        x = self.maxpool(x)
+
+        x = self.layer1(x)
+        x = self.layer2(x)
+        x = self.layer3(x)
+        x = self.layer4(x)
+
+        x = self.avgpool(x)
+        x = x.view(x.shape[0], -1)
+        x = self.fc(x)
+        return x

style.css

@@ -0,0 +1,83 @@
+* {
+	box-sizing: border-box;
+}
+
+body {
+	font-family: 'Source Sans Pro', sans-serif;
+	font-size: 16px;
+}
+
+.container {
+	width: 100%;
+	margin: 0 auto;
+}
+
+.title {
+	font-size: 24px !important;
+	font-weight: 600 !important;
+	letter-spacing: 0em;
+	text-align: center;
+	color: #374159 !important;
+}
+
+.subtitle {
+	font-size: 24px !important;
+	font-style: italic;
+	font-weight: 400 !important;
+	letter-spacing: 0em;
+	text-align: center;
+	color: #1d652a !important;
+	padding-bottom: 0.5em;
+}
+
+.overview-heading {
+	font-size: 24px !important;
+	font-weight: 600 !important;
+	letter-spacing: 0em;
+	text-align: left;
+}
+
+.overview-content {
+	font-size: 14px !important;
+	font-weight: 400 !important;
+	line-height: 30px !important;
+	letter-spacing: 0em;
+	text-align: left;
+}
+
+.content-image {
+	width: 100% !important;
+	height: auto !important;
+}
+
+.vl {
+	border-left: 5px solid #1d652a;
+	padding-left: 20px;
+	color: #1d652a !important;
+}
+
+.grid-container {
+	display: grid;
+	grid-template-columns: 1fr 2fr;
+	gap: 20px;
+	align-items: flex-start;
+	margin-bottom: 0.7em;
+}
+
+.grid-container:nth-child(2) {
+	align-items: center;
+}
+
+@media screen and (max-width: 768px) {
+	.container {
+		width: 90%;
+	}
+
+	.grid-container {
+		display: block;
+	}
+
+	.overview-heading {
+		font-size: 18px !important;
+	}
+}