Upload 164 files

.gitattributes

@@ -77,3 +77,47 @@ SlimFace/data/reference_data/images/Tom[[:space:]]Cruise.jpg filter=lfs diff=lfs
 SlimFace/data/reference_data/images/Vijay[[:space:]]Deverakonda.jpg filter=lfs diff=lfs merge=lfs -text
 SlimFace/data/reference_data/images/Virat[[:space:]]Kohli.jpg filter=lfs diff=lfs merge=lfs -text
 SlimFace/data/reference_data/images/Zac[[:space:]]Efron.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Akshay[[:space:]]Kumar.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Alexandra[[:space:]]Daddario.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Alia[[:space:]]Bhatt.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Amitabh[[:space:]]Bachchan.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Andy[[:space:]]Samberg.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Anushka[[:space:]]Sharma.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Billie[[:space:]]Eilish.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Brad[[:space:]]Pitt.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Camila[[:space:]]Cabello.png filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Charlize[[:space:]]Theron.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Claire[[:space:]]Holt.png filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Courtney[[:space:]]Cox.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Dwayne[[:space:]]Johnson.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Elizabeth[[:space:]]Olsen.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Ellen[[:space:]]Degeneres.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Henry[[:space:]]Cavill.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Hrithik[[:space:]]Roshan.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Hugh[[:space:]]Jackman.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Jessica[[:space:]]Alba.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Kashyap.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Lisa[[:space:]]Kudrow.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Margot[[:space:]]Robbie.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Natalie[[:space:]]Portman.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Priyanka[[:space:]]Chopra.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Roger[[:space:]]Federer.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Tom[[:space:]]Cruise.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Vijay[[:space:]]Deverakonda.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Virat[[:space:]]Kohli.jpg filter=lfs diff=lfs merge=lfs -text
+data/reference_data/images/Zac[[:space:]]Efron.jpg filter=lfs diff=lfs merge=lfs -text
+src/slimface/models/detection_models/mtcnn_pytorch/caffe_models/det2.caffemodel filter=lfs diff=lfs merge=lfs -text
+src/slimface/models/detection_models/mtcnn_pytorch/caffe_models/det3.caffemodel filter=lfs diff=lfs merge=lfs -text
+src/slimface/models/detection_models/mtcnn_pytorch/caffe_models/det4.caffemodel filter=lfs diff=lfs merge=lfs -text
+src/slimface/models/detection_models/mtcnn_pytorch/images/example.png filter=lfs diff=lfs merge=lfs -text
+src/slimface/models/detection_models/mtcnn_pytorch/images/office2.jpg filter=lfs diff=lfs merge=lfs -text
+src/slimface/models/detection_models/mtcnn_pytorch/images/office4.jpg filter=lfs diff=lfs merge=lfs -text
+tests/test_images/Cate[[:space:]]Blanchett.jpg filter=lfs diff=lfs merge=lfs -text
+tests/test_images/Daniel[[:space:]]Day-Lewis.jpg filter=lfs diff=lfs merge=lfs -text
+tests/test_images/dont_know.jpg filter=lfs diff=lfs merge=lfs -text
+tests/test_images/Elon_Musk.jpg filter=lfs diff=lfs merge=lfs -text
+tests/test_images/Gal[[:space:]]Gado.jpg filter=lfs diff=lfs merge=lfs -text
+tests/test_images/Kate[[:space:]]Winslet.jpg filter=lfs diff=lfs merge=lfs -text
+tests/test_images/Tom[[:space:]]Cruise.jpg filter=lfs diff=lfs merge=lfs -text
+tests/test_images/Tom[[:space:]]Hanks.jpg filter=lfs diff=lfs merge=lfs -text
+tests/test_images/Viola[[:space:]]Davis.jpg filter=lfs diff=lfs merge=lfs -text

.python-version

@@ -0,0 +1 @@
+3.11.11

LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2025 Danh Tran
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

README.md

@@ -1,13 +1,128 @@
----
-title: Slimface Demonstration
-emoji: 👦
-colorFrom: blue
-colorTo: purple
-sdk: gradio
-sdk_version: 5.38.2
-app_file: SlimFace/apps/gradio_app.py
-pinned: true
-license: mit
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# SlimFace: Slim Face Recognition
+
+> ## Credits and Citation
+>
+> ℹ️ This project is based on the [![Built on edgeface](https://img.shields.io/badge/Built%20on-otroshi%2Fedgeface-blue?style=flat&logo=github)](https://github.com/otroshi/edgeface) by [![Hatef Otroshi](https://img.shields.io/badge/GitHub-Hatef_Otroshi-blue?style=flat&logo=github)](https://github.com/otroshi), and includes our own bug fixes and enhancements.
+>
+> If this project is helpful for your research, please consider citing the original paper:
+>
+> **Edgeface: Efficient face recognition model for edge devices**  
+> *George, Anjith and Ecabert, Christophe and Shahreza, Hatef Otroshi and Kotwal, Ketan and Marcel, Sebastien*  
+> *IEEE Transactions on Biometrics, Behavior, and Identity Science (2024)*
+>
+> **If you use this work in your research, please cite the original paper:**
+> ```bibtex
+> @article{edgeface,
+>   title={Edgeface: Efficient face recognition model for edge devices},
+>   author={George, Anjith and Ecabert, Christophe and Shahreza, Hatef Otroshi and Kotwal, Ketan and Marcel, Sebastien},
+>   journal={IEEE Transactions on Biometrics, Behavior, and Identity Science},
+>   year={2024}
+> }
+> ```
+
+
+## Usage
+### Clone Repositories
+```bash
+# Clone the repository
+git clone https://github.com/danhtran2mind/SlimFace
+
+# Navigate into the newly created 'slimface' directory.
+cd SlimFace
+```
+### Install Dependencies
+**If Open-CV (CV2) does not work, run below CLI**
+```bash
+sudo apt update
+sudo apt install -y libglib2.0-0
+sudo apt install -y libgl1-mesa-dev
+```
+### Default install Dependencies
+```bash
+pip install -r requirements/requirements.txt
+```
+### Other install Dependencies
+- For My Compatible
+```bash
+pip install -r requirements/requirements_compatible.txt
+```
+- For `End2end Inference`
+```bash
+pip install -r requirements/requirements_inference.txt
+```
+### Download Model Checkpoints
+```bash
+python scripts/download_ckpts.py
+```
+### Setup Third Party
+```bash
+python scripts/setup_third_party.py
+```
+## Data Preparation
+
+## Pre-trained Model preparation
+For detailed instructions on how to process and manage your data effectively, refer to the [Full guide for data processing](./docs/data_processing.md).
+
+This is fast usage for dataset preparation
+```bash
+python scripts/process_dataset.py
+```
+## Training
+
+1. Configure the default settings for Accelerate:
+```bash
+accelerate config default
+```
+
+2. Launch the training script using Accelerate:
+```bash
+accelerate launch src/slimface/training/accelerate_train.py
+```
+
+For additional help, you can refer to the [Training Documentation](./docs/training/training_docs.md) for more details.
+
+### Inference
+#### Create Reference Images Data at `data/reference_data/images`
+For each class, you store an image in `data/reference_data/images` folder which are maped with `index_to_class_mapping.json`.
+
+The structure like:
+```markdown
+data/reference_data/images/
+├── 'Robert Downey Jr.jpg'
+├── 'Tom Cruise.jpg'
+└── ...
+```
+
+
+### Create Reference Dictionary from `index_to_class_mapping.json`
+
+#### Steps
+1. Place `index_to_class_mapping.json` in the `ckpts` folder.
+2. Ensure reference images are in `data/reference_data/images`. Missing images will be set to `""` in `reference_image_data.json` (default in `data/reference_data` folder).
+3. Run one of the following commands:
+
+#### Commands
+- **Default** (Output: `data/reference_data/reference_image_data.json`):
+  ```bash
+  python scripts/create_reference_image_path.py
+  ```
+- **Custom Paths**:
+  ```bash
+  python scripts/create_reference_image_path.py \
+      --input <path_to_index_to_class_mapping.json> \
+      --output <path_to_tests/reference_image_data.json>
+  ```
+
+#### Manual Option
+Edit `reference_image_data.json` directly to add image paths as dictionary values.
+
+## Demostration
+```bash
+python apps/gradio_app.py
+```
+
+https://huggingface.co/spaces/danhtran2mind/SlimFace-demo
+
+## Project Description
+
+This repository is trained from [![GitHub Repo](https://img.shields.io/badge/GitHub-danhtran2mind%2Fedgeface-blue?style=flat)](https://github.com/danhtran2mind/edgeface), a fork of [![GitHub Repo](https://img.shields.io/badge/GitHub-otroshi%2Fedgeface-blue?style=flat)](https://github.com/otroshi/edgeface), with numerous bug fixes and rewritten code for improved performance and stability.

apps/gradio_app.py

@@ -0,0 +1,116 @@
+import gradio as gr
+from PIL import Image
+from gradio_app.inference import run_inference
+from gradio_app.components import (
+    CONTENT_DESCRIPTION, CONTENT_IN, CONTENT_OUT,
+     list_reference_files, list_mapping_files,
+      list_classifier_files, list_edgeface_files
+)
+
+def create_image_input_column():
+    """Create the column for image input and output display."""
+    with gr.Column():
+        image_input = gr.Image(type="pil", label="Upload Image")
+        output = gr.HTML(label="Inference Results", elem_classes=["results-container"])
+    return image_input, output
+
+def create_model_files_column():
+    """Create the column for model file selection."""
+    with gr.Column():
+        with gr.Group(elem_classes=["section-group"]):
+            gr.Markdown("### Model Files", elem_classes=["section-title"])
+            ref_dict = gr.Dropdown(
+                choices=["Select a file"] + list_reference_files(),
+                label="Reference Dict JSON",
+                value="data/reference_data/reference_image_data.json"
+            )
+            index_map = gr.Dropdown(
+                choices=["Select a file"] + list_mapping_files(),
+                label="Index to Class Mapping JSON",
+                value="ckpts/index_to_class_mapping.json"
+            )
+            classifier_model = gr.Dropdown(
+                choices=["Select a file"] + list_classifier_files(),
+                label="Classifier Model (.pth)",
+                value="ckpts/SlimFace_efficientnet_b3_full_model.pth"
+            )
+            edgeface_model = gr.Dropdown(
+                choices=["Select a file"] + list_edgeface_files(),
+                label="EdgeFace Model (.pt)",
+                value="ckpts/idiap/edgeface_s_gamma_05.pt"
+            )
+    return ref_dict, index_map, classifier_model, edgeface_model
+
+def create_settings_column():
+    """Create the column for advanced settings."""
+    with gr.Column():
+        with gr.Group(elem_classes=["section-group"]):
+            gr.Markdown("### Advanced Settings", elem_classes=["section-title"])
+            algorithm = gr.Dropdown(
+                choices=["yolo", "mtcnn", "retinaface"],
+                label="Detection Algorithm",
+                value="yolo"
+            )
+            accelerator = gr.Dropdown(
+                choices=["auto", "cpu", "cuda", "mps"],
+                label="Accelerator",
+                value="auto"
+            )
+            resolution = gr.Slider(
+                minimum=128,
+                maximum=512,
+                step=32,
+                label="Image Resolution",
+                value=300
+            )
+            similarity_threshold = gr.Slider(
+                minimum=0.1,
+                maximum=1.0,
+                step=0.05,
+                label="Similarity Threshold",
+                value=0.3
+            )
+    return algorithm, accelerator, resolution, similarity_threshold
+
+def create_interface():
+    """Create the Gradio interface for SlimFace."""
+    with gr.Blocks(css="gradio_app/static/styles.css", theme=gr.themes.Soft()) as demo:
+        gr.Markdown("# SlimFace Demonstration")
+        gr.Markdown(CONTENT_DESCRIPTION)
+        gr.HTML(CONTENT_IN)
+        
+        with gr.Row():
+            image_input, output = create_image_input_column()
+            ref_dict, index_map, classifier_model, edgeface_model = create_model_files_column()
+        
+        with gr.Row():
+            algorithm, accelerator, resolution, similarity_threshold = create_settings_column()
+        
+        with gr.Row():
+            submit_btn = gr.Button("Run Inference", variant="primary", elem_classes=["centered-button"])
+        
+        submit_btn.click(
+            fn=run_inference,
+            inputs=[
+                image_input,
+                ref_dict,
+                index_map,
+                classifier_model,
+                edgeface_model,
+                algorithm,
+                accelerator,
+                resolution,
+                similarity_threshold
+            ],
+            outputs=output
+        )
+        gr.Markdown(CONTENT_OUT)
+    return demo
+
+def main():
+    """Launch the Gradio interface."""
+    demo = create_interface()
+    demo.launch()
+
+if __name__ == "__main__":
+    main()

apps/gradio_app/components.py

@@ -0,0 +1,135 @@
+import os
+
+# File listing functions
+def list_reference_files():
+    ref_dir = "data/reference_data/"
+    try:
+        files = [os.path.join(ref_dir, f) for f in os.listdir(ref_dir) if f.endswith(".json")]
+        return files if files else ["No .json files found in data/reference_data/"]
+    except FileNotFoundError:
+        return ["Directory data/reference_data/ not found"]
+    except Exception as e:
+        return [f"Error listing files: {str(e)}"]
+
+def list_mapping_files():
+    map_dir = "ckpts/"
+    try:
+        files = [os.path.join(map_dir, f) for f in os.listdir(map_dir) if f.endswith(".json")]
+        return files if files else ["No .json files found in ckpts/"]
+    except FileNotFoundError:
+        return ["Directory ckpts/ not found"]
+    except Exception as e:
+        return [f"Error listing files: {str(e)}"]
+
+def list_classifier_files():
+    clf_dir = "ckpts/"
+    try:
+        files = [os.path.join(clf_dir, f) for f in os.listdir(clf_dir) if f.endswith(".pth")]
+        return files if files else ["No .pth files found in ckpts/"]
+    except FileNotFoundError:
+        return ["Directory ckpts/ not found"]
+    except Exception as e:
+        return [f"Error listing files: {str(e)}"]
+
+def list_edgeface_files():
+    ef_dir = "ckpts/idiap/"
+    try:
+        files = [os.path.join(ef_dir, f) for f in os.listdir(ef_dir) if f.endswith(".pt")]
+        return files if files else ["No .pt files found in ckpts/idiap/"]
+    except FileNotFoundError:
+        return ["Directory ckpts/idiap/ not found"]
+    except Exception as e:
+        return [f"Error listing files: {str(e)}"]
+
+CONTENT_DESCRIPTION = """
+**SlimFace: Advanced Face Classification with TorchVision Backbones**
+"""
+CONTENT_IN = """
+<style>
+    body {
+        font-family: Arial, sans-serif;
+        line-height: 1.6;
+        margin: 0; /* Remove default margin for full-width */
+        padding: 20px; /* Adjust padding for content spacing */
+        color: #333;
+        width: 100%; /* Ensure body takes full width */
+        box-sizing: border-box; /* Include padding in width calculation */
+    }
+    .attribution {
+        background-color: #f9f9f9;
+        padding: 20px;
+        border-radius: 8px;
+        box-shadow: 0 2px 4px rgba(0, 0, 0, 0.1);
+    }
+    .quote-container {
+        border-left: 5px solid #007bff;
+        padding-left: 15px;
+        margin-bottom: 15px;
+        font-style: italic;
+    }
+    .attribution p {
+        margin: 10px 0;
+    }
+    .badge {
+        display: inline-block;
+        border-radius: 4px;
+        text-decoration: none;
+        font-size: 14px;
+        transition: background-color 0.3s;
+    }
+    .badge:hover {
+        background-color: #0056b3;
+    }
+    .badge img {
+        vertical-align: middle;
+        margin-right: 5px;
+    }
+    .source {
+        color: #555;
+    }
+</style>
+    <div class="quote-container">
+        <p>
+            This project leverages code from 
+            <a class="badge" href="https://github.com/otroshi/edgeface">
+                <img src="https://img.shields.io/badge/Built%20on-otroshi%2Fedgeface-blue?style=flat&logo=github" alt="Built on edgeface">
+            </a>
+            by 
+            <a class="badge" href="https://github.com/otroshi">
+                <img src="https://img.shields.io/badge/GitHub-Hatef_Otroshi-blue?style=flat&logo=github" alt="Hatef Otroshi">
+            </a>, 
+            with our own bug fixes and enhancements available at 
+            <a class="badge" href="https://github.com/danhtran2mind/edgeface/tree/main/face_alignment">
+                <img src="https://img.shields.io/badge/GitHub-danhtran2mind%2Fedgeface-blue?style=flat&logo=github" alt="Edgeface Enhancements">
+            </a>.
+        </p>
+    </div>
+    <p class="source">
+        For more information, you can follow below:<br>
+        Source code: 
+        <a class="badge" href="https://github.com/danhtran2mind/SlimFace">
+            <img src="https://img.shields.io/badge/GitHub-danhtran2mind%2FSlimFace-blue?style=flat" alt="GitHub Repo">
+            ,
+        </a>
+        Author: 
+        <a class="badge" href="https://github.com/danhtran2mind">
+            <img src="https://img.shields.io/badge/GitHub-danhtran2mind-blue?style=flat" alt="GitHub Profile">
+            ,
+        </a>
+        PyTorch Docs: 
+        <a class="badge" href="https://docs.pytorch.org/vision/main/models.html">
+            <img src="https://img.shields.io/badge/PyTorch-Pretrain%20Model%20Docs-blue?style=flat" alt="PyTorch Docs">
+        </a>
+    </p>
+"""
+
+CONTENT_OUT = """
+## More Information about SlimFace
+
+SlimFace empowers developers to build high-accuracy face classification models using transfer learning, leveraging TorchVision's powerful pre-trained architectures. 🌟 It provides a flexible, efficient, and scalable solution for facial recognition, delivering top-tier performance for custom applications.
+
+**Supported Architectures:**  
+- **EfficientNet**: B0-B7 and V2 (Small, Medium, Large) for balanced performance and efficiency. 📸  
+- **RegNet**: X/Y series (400MF to 128GF) for optimized computation across diverse hardware. 💻  
+- **Vision Transformers (ViT)**: B_16, B_32, H_14, L_16, L_32 for cutting-edge feature extraction. 🚀  
+"""

apps/gradio_app/inference.py

@@ -0,0 +1,93 @@
+import os
+import sys
+from PIL import Image
+
+# Append the path to the inference script's directory
+sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), '..', '..', 'src', 'slimface', 'inference')))
+
+from end2end_inference import inference_and_confirm
+
+def run_inference(image, reference_dict_path, index_to_class_mapping_path, model_path,
+                 edgeface_model_path="ckpts/idiap/edgeface_base.pt", 
+                 algorithm="yolo", accelerator="auto", resolution=224, similarity_threshold=0.6):
+    
+    # Validate image input
+    if image is None:
+        return '<div class="error-message">Error: No image provided. Please upload an image.</div>'
+
+    # Define temporary image path
+    temp_image_path = os.path.join(os.path.dirname(__file__), "temp_data", "temp_image.jpg")
+    os.makedirs(os.path.dirname(temp_image_path), exist_ok=True)
+    
+    # Save the image
+    try:
+        image.save(temp_image_path)
+    except Exception as e:
+        return f'<div class="error-message">Error saving image: {str(e)}</div>'
+
+    # Create args object to mimic command-line arguments
+    class Args:
+        def __init__(self):
+            self.unknown_image_path = temp_image_path
+            self.reference_dict_path = reference_dict_path.name if hasattr(reference_dict_path, 'name') else reference_dict_path
+            self.index_to_class_mapping_path = index_to_class_mapping_path.name if hasattr(index_to_class_mapping_path, 'name') else index_to_class_mapping_path
+            self.model_path = model_path.name if hasattr(model_path, 'name') else model_path
+            self.edgeface_model_path = edgeface_model_path.name if hasattr(edgeface_model_path, 'name') else edgeface_model_path
+            self.algorithm = algorithm
+            self.accelerator = accelerator
+            self.resolution = resolution
+            self.similarity_threshold = similarity_threshold
+
+    args = Args()
+
+    # Validate inputs
+    if not all([args.reference_dict_path, args.index_to_class_mapping_path, args.model_path]):
+        return '<div class="error-message">Error: Please provide all required files (reference dict, index-to-class mapping, and model).</div>'
+
+    try:
+        # Call the inference function from end2end_inference.py
+        results = inference_and_confirm(args)
+        
+        # Format output as HTML for Gradio
+        output = '<div class="results-container">'
+        output += '<h2 class="result-title">Inference Results</h2>'
+        
+        if not results:
+            output += '<div class="error-message">No results returned from inference.</div>'
+        else:
+            for idx, result in enumerate(results, 1):
+                output += '<div class="result-card">'
+                output += f'<h3 class="result-title">Result {idx}</h3>'
+                
+                # Person Name
+                person_name = result.get('predicted_class', 'N/A')
+                output += f'<div class="result-item"><span class="label">Person Name</span><span class="value">{person_name}</span></div>'
+                
+                # Confidence
+                confidence = result.get('confidence', 'N/A')
+                confidence_str = f'{confidence:.4f}' if isinstance(confidence, (int, float)) else 'N/A'
+                output += f'<div class="result-item"><span class="label">Confidence</span><span class="value">{confidence_str}</span></div>'
+                
+                # Similarity with Reference Image
+                similarity = result.get('similarity', 'N/A')
+                similarity_str = f'{similarity:.4f}' if isinstance(similarity, (int, float)) else 'N/A'
+                output += f'<div class="result-item"><span class="label">Similarity with<br>Reference Image</span><span class="value">{similarity_str}</span></div>'
+                
+                # Confirmed Person
+                confirmed = result.get('confirmed', 'N/A')
+                confirmed_class = 'confirmed-true' if confirmed is True else 'confirmed-false' if confirmed is False else ''
+                confirmed_str = str(confirmed) if confirmed is not None else 'N/A'
+                output += f'<div class="result-item"><span class="label">Confirmed Person</span><span class="value {confirmed_class}">{confirmed_str}</span></div>'
+                
+                output += '</div>'
+        
+        output += '</div>'
+        return output
+    
+    except Exception as e:
+        return f'<div class="error-message">Error during inference: {str(e)}</div>'
+    
+    finally:
+        # Clean up temporary image
+        if os.path.exists(temp_image_path):
+            os.remove(temp_image_path)

apps/gradio_app/static/script.js

@@ -0,0 +1,3 @@
+// Placeholder for future JavaScript functionality
+// Currently, no JavaScript is required for the Gradio app as interactions are handled by Gradio
+console.log("SlimFace Gradio App JavaScript loaded");

apps/gradio_app/static/style.css

@@ -0,0 +1,229 @@
+body {
+    font-family: 'Inter', -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, sans-serif;
+    background: linear-gradient(145deg, #e2e8f0 0%, #b8c6db 100%);
+    margin: 0;
+    padding: 0;
+    min-height: 100vh;
+    color: #1a202c;
+}
+
+.gradio-container {
+    max-width: 1280px;
+    margin: 0 auto;
+    padding: 2.5rem 1.5rem;
+    box-sizing: border-box;
+}
+
+h1 {
+    color: #1a202c;
+    font-size: 2.75rem;
+    font-weight: 800;
+    text-align: center;
+    margin-bottom: 2.5rem;
+    letter-spacing: -0.025em;
+    background: linear-gradient(to right, #2b6cb0, #4a90e2);
+    -webkit-background-clip: text;
+    -webkit-text-fill-color: transparent;
+}
+
+.section-title {
+    color: #1a202c;
+    font-size: 1.5rem;
+    font-weight: 700;
+    margin-bottom: 1rem;
+    border-bottom: 2px solid #4a90e2;
+    padding-bottom: 0.5rem;
+    letter-spacing: -0.015em;
+}
+
+.section-group {
+    background: rgba(255, 255, 255, 0.95);
+    border-radius: 0.5rem;
+    padding: 1.5rem;
+    border: 1px solid rgba(226, 232, 240, 0.5);
+    box-shadow: 0 2px 4px rgba(0, 0, 0, 0.05);
+}
+
+.results-container {
+    display: flex;
+    flex-direction: column;
+    gap: 1.75rem;
+    padding: 2rem;
+    background: rgba(255, 255, 255, 0.95);
+    border-radius: 1.25rem;
+    box-shadow: 0 10px 20px rgba(0, 0, 0, 0.15), 0 4px 6px rgba(0, 0, 0, 0.1);
+    border: 1px solid rgba(226, 232, 240, 0.5);
+    backdrop-filter: blur(8px);
+}
+
+.result-card {
+    background: linear-gradient(145deg, #f7fafc, #edf2f7);
+    border-radius: 1rem;
+    padding: 2.25rem;
+    box-shadow: 0 6px 12px rgba(0, 0, 0, 0.1);
+    transition: transform 0.3s ease, box-shadow 0.3s ease, background 0.3s ease;
+    position: relative;
+    overflow: hidden;
+}
+
+.result-card:hover {
+    transform: translateY(-5px);
+    box-shadow: 0 10px 24px rgba(0, 0, 0, 0.15);
+    background: linear-gradient(145deg, #ffffff, #e6eefa);
+}
+
+.result-card::before {
+    content: '';
+    position: absolute;
+    top: 0;
+    left: 0;
+    width: 100%;
+    height: 4px;
+    background: linear-gradient(to right, #4a90e2, #63b3ed);
+    transition: height 0.3s ease;
+}
+
+.result-card:hover::before {
+    height: 8px;
+}
+
+.result-title {
+    color: #1a202c;
+    font-size: 1.875rem;
+    font-weight: 700;
+    margin-bottom: 1.5rem;
+    border-bottom: 3px solid #4a90e2;
+    padding-bottom: 0.75rem;
+    letter-spacing: -0.015em;
+}
+
+.result-item {
+    display: flex;
+    justify-content: space-between;
+    align-items: center;
+    margin: 1rem 0;
+    font-size: 1.125rem;
+    color: #2d3748;
+    line-height: 1.6;
+}
+
+.label {
+    font-weight: 600;
+    color: #2b6cb0;
+    text-align: left;
+    text-transform: uppercase;
+    font-size: 0.95rem;
+    letter-spacing: 0.05em;
+    flex: 0 0 auto;
+}
+
+.value {
+    color: #1a202c;
+    font-weight: 500;
+    text-align: right;
+    flex: 0 0 auto;
+}
+
+.value.confirmed-true {
+    color: #2f855a;
+    font-weight: 600;
+    background: #c6f6d5;
+    padding: 0.25rem 0.5rem;
+    border-radius: 0.375rem;
+}
+
+.value.confirmed-false {
+    color: #c53030;
+    font-weight: 600;
+    background: #fed7d7;
+    padding: 0.25rem 0.5rem;
+    border-radius: 0.375rem;
+}
+
+.error-message {
+    background: #fef2f2;
+    color: #9b2c2c;
+    padding: 1.75rem;
+    border-radius: 0.875rem;
+    margin: 1.25rem 0;
+    font-size: 1.125rem;
+    font-weight: 500;
+    border: 1px solid #e53e3e;
+    box-shadow: 0 2px 8px rgba(0, 0, 0, 0.1);
+}
+
+.centered-button {
+    display: block;
+    margin: 1rem auto;
+    background: #4a90e2;
+    color: white;
+    padding: 0.75rem 1.5rem;
+    border-radius: 0.5rem;
+    border: none;
+    font-size: 1rem;
+    font-weight: 600;
+    cursor: pointer;
+    transition: background 0.3s ease;
+    position: relative;
+    padding-left: 2.5rem;
+    width: 30%;
+}
+
+.centered-button:hover {
+    background: #2b6cb0;
+}
+
+.centered-button::after {
+    content: '🤔';
+    position: absolute;
+    left: 0.75rem;
+    top: 50%;
+    transform: translateY(-50%);
+    font-size: 1.2rem;
+}
+
+@media (max-width: 768px) {
+    .gradio-container {
+        padding: 1.5rem;
+    }
+    
+    h1 {
+        font-size: 2rem;
+    }
+    
+    .results-container {
+        padding: 1.5rem;
+    }
+    
+    .result-card {
+        padding: 1.5rem;
+    }
+    
+    .result-title {
+        font-size: 1.5rem;
+    }
+    
+    .result-item {
+        font-size: 1rem;
+        flex-direction: column;
+        align-items: flex-start;
+        gap: 0.5rem;
+    }
+    
+    .label, .value {
+        text-align: left;
+    }
+    
+    .section-title {
+        font-size: 1.25rem;
+    }
+    
+    .section-group {
+        padding: 1rem;
+    }
+    
+    .centered-button {
+        padding: 0.5rem 1rem;
+        font-size: 0.9rem;
+    }
+}

assets/comparision.md

@@ -0,0 +1,11 @@
+slim_face_vit_b_16
+    Train loss: 0.0074, Train acc: 0.9980, Val loss: 0.2179, Val acc: 0.9336
+    
+efficientnet_b3
+    Train loss: 0.0014, Train acc: 1.0000, Val loss: 0.1931, Val acc: 0.9427
+
+efficientnet_v2_s
+    Train loss: 0.0016, Train acc: 1.0000, Val loss: 0.2374, Val acc: 0.9375
+
+regnet_y_800mf
+    Train loss: 0.0033, Train acc: 0.9997, Val loss: 0.3766, Val acc: 0.8906

ckpts/.gitignore

@@ -0,0 +1,4 @@
+# Ignore everything in this directory
+*
+# Except this .gitignore file
+!.gitignore

configs/accelerate_config.yaml

@@ -0,0 +1,6 @@
+compute_environment: LOCAL_MACHINE
+distributed_type: FSDP
+num_processes: 4
+mixed_precision: fp16
+fsdp_config:
+  fsdp_offload_params: true

configs/image_classification_models_config.yaml

@@ -0,0 +1,249 @@
+# For more details on models, see https://pytorch.org/vision/main/models.html
+
+# EfficientNet models: Designed for efficiency with compound scaling of depth, width, and resolution.
+# These models balance accuracy and computational efficiency, ideal for resource-constrained environments.
+efficientnet_b0:
+  metrics:
+    Acc@1: 77.692  # Top-1 accuracy on ImageNet
+    Acc@5: 93.532  # Top-5 accuracy on ImageNet
+    GFLOPS: 0.39   # Computational complexity
+    Params: 5.3M   # Number of parameters
+  model_fn: models.efficientnet_b0
+  resolution: 224  # Input image resolution
+  weights: models.EfficientNet_B0_Weights.IMAGENET1K_V1  # Pretrained weights on ImageNet
+
+efficientnet_b1:
+  metrics:
+    Acc@1: 78.642
+    Acc@5: 94.186
+    GFLOPS: 0.69
+    Params: 7.8M
+  model_fn: models.efficientnet_b1
+  resolution: 240
+  weights: models.EfficientNet_B1_Weights.IMAGENET1K_V1
+
+efficientnet_b2:
+  metrics:
+    Acc@1: 80.608
+    Acc@5: 95.31
+    GFLOPS: 1.09
+    Params: 9.1M
+  model_fn: models.efficientnet_b2
+  resolution: 260
+  weights: models.EfficientNet_B2_Weights.IMAGENET1K_V1
+
+efficientnet_b3:
+  metrics:
+    Acc@1: 82.008
+    Acc@5: 96.054
+    GFLOPS: 1.83
+    Params: 12.2M
+  model_fn: models.efficientnet_b3
+  resolution: 300
+  weights: models.EfficientNet_B3_Weights.IMAGENET1K_V1
+
+efficientnet_b4:
+  metrics:
+    Acc@1: 83.384
+    Acc@5: 96.594
+    GFLOPS: 4.39
+    Params: 19.3M
+  model_fn: models.efficientnet_b4
+  resolution: 380
+  weights: models.EfficientNet_B4_Weights.IMAGENET1K_V1
+
+efficientnet_b5:
+  metrics:
+    Acc@1: 83.444
+    Acc@5: 96.628
+    GFLOPS: 10.27
+    Params: 30.4M
+  model_fn: models.efficientnet_b5
+  resolution: 456
+  weights: models.EfficientNet_B5_Weights.IMAGENET1K_V1
+
+efficientnet_b6:
+  metrics:
+    Acc@1: 84.008
+    Acc@5: 96.916
+    GFLOPS: 19.07
+    Params: 43.0M
+  model_fn: models.efficientnet_b6
+  resolution: 528
+  weights: models.EfficientNet_B6_Weights.IMAGENET1K_V1
+
+efficientnet_b7:
+  metrics:
+    Acc@1: 84.122
+    Acc@5: 96.908
+    GFLOPS: 37.75
+    Params: 66.3M
+  model_fn: models.efficientnet_b7
+  resolution: 600
+  weights: models.EfficientNet_B7_Weights.IMAGENET1K_V1
+
+# EfficientNet V2 models: Improved training efficiency and performance over V1.
+# These models use progressive learning and optimized scaling for better accuracy.
+efficientnet_v2_l:
+  metrics:
+    Acc@1: 85.808
+    Acc@5: 97.788
+    GFLOPS: 56.08
+    Params: 118.5M
+  model_fn: models.efficientnet_v2_l
+  resolution: 480
+  weights: models.EfficientNet_V2_L_Weights.IMAGENET1K_V1
+
+efficientnet_v2_m:
+  metrics:
+    Acc@1: 85.112
+    Acc@5: 97.156
+    GFLOPS: 24.58
+    Params: 54.1M
+  model_fn: models.efficientnet_v2_m
+  resolution: 480
+  weights: models.EfficientNet_V2_M_Weights.IMAGENET1K_V1
+
+efficientnet_v2_s:
+  metrics:
+    Acc@1: 84.228
+    Acc@5: 96.878
+    GFLOPS: 8.37
+    Params: 21.5M
+  model_fn: models.efficientnet_v2_s
+  resolution: 384
+  weights: models.EfficientNet_V2_S_Weights.IMAGENET1K_V1
+
+# RegNet models: Designed for scalability and efficiency with a focus on network design.
+# These models optimize for both accuracy and computational efficiency.
+regnet_y_128gf:
+  metrics:
+    Acc@1: 86.068  # High accuracy but computationally expensive
+    Acc@5: 97.844
+    GFLOPS: 127.52
+    Params: 644.8M
+  model_fn: models.regnet_y_128gf
+  resolution: 224
+  weights: models.RegNet_Y_128GF_Weights.IMAGENET1K_SWAG_LINEAR_V1
+
+regnet_y_16gf:
+  metrics:
+    Acc@1: 82.886
+    Acc@5: 96.328
+    GFLOPS: 15.91
+    Params: 83.6M
+  model_fn: models.regnet_y_16gf
+  resolution: 224
+  weights: models.RegNet_Y_16GF_Weights.IMAGENET1K_V2
+
+regnet_y_1_6gf:
+  metrics:
+    Acc@1: 80.876
+    Acc@5: 95.444
+    GFLOPS: 1.61
+    Params: 11.2M
+  model_fn: models.regnet_y_1_6gf
+  resolution: 224
+  weights: models.RegNet_Y_1_6GF_Weights.IMAGENET1K_V2
+
+regnet_y_32gf:
+  metrics:
+    Acc@1: 83.368
+    Acc@5: 96.498
+    GFLOPS: 32.28
+    Params: 145.0M
+  model_fn: models.regnet_y_32gf
+  resolution: 224
+  weights: models.RegNet_Y_32GF_Weights.IMAGENET1K_V2
+
+regnet_y_3_2gf:
+  metrics:
+    Acc@1: 81.982
+    Acc@5: 95.972
+    GFLOPS: 3.18
+    Params: 19.4M
+  model_fn: models.regnet_y_3_2gf
+  resolution: 224
+  weights: models.RegNet_Y_3_2GF_Weights.IMAGENET1K_V2
+
+regnet_y_400mf:
+  metrics:
+    Acc@1: 75.804
+    Acc@5: 92.742
+    GFLOPS: 0.4
+    Params: 4.3M
+  model_fn: models.regnet_y_400mf
+  resolution: 224
+  weights: models.RegNet_Y_400MF_Weights.IMAGENET1K_V2
+
+regnet_y_800mf:
+  metrics:
+    Acc@1: 78.828
+    Acc@5: 94.502
+    GFLOPS: 0.83
+    Params: 6.4M
+  model_fn: models.regnet_y_800mf
+  resolution: 224
+  weights: models.RegNet_Y_800MF_Weights.IMAGENET1K_V2
+
+regnet_y_8gf:
+  metrics:
+    Acc@1: 82.828
+    Acc@5: 96.33
+    GFLOPS: 8.47
+    Params: 39.4M
+  model_fn: models.regnet_y_8gf
+  resolution: 224
+  weights: models.RegNet_Y_8GF_Weights.IMAGENET1K_V2
+
+# Vision Transformer (ViT) models: Transformer-based architecture for image classification.
+# These models excel in capturing long-range dependencies but require significant compute for larger variants.
+vit_b_16:
+  metrics:
+    Acc@1: 81.072  # Base ViT model with balanced accuracy and efficiency
+    Acc@5: 95.318
+    GFLOPS: 17.56
+    Params: 86.6M
+  model_fn: models.vit_b_16
+  resolution: 224
+  weights: models.ViT_B_16_Weights.IMAGENET1K_V1
+
+vit_b_32:
+  metrics:
+    Acc@1: 75.912  # Smaller patch size version of ViT, lower accuracy but fewer computations
+    Acc@5: 92.466
+    GFLOPS: 4.41
+    Params: 88.2M
+  model_fn: models.vit_b_32
+  resolution: 224
+  weights: models.ViT_B_32_Weights.IMAGENET1K_V1
+
+vit_h_14:
+  metrics:
+    Acc@1: 88.552  # High-performance ViT model with very high accuracy and computational cost
+    Acc@5: 98.694
+    GFLOPS: 1016.72
+    Params: 633.5M
+  model_fn: models.vit_h_14
+  resolution: 224
+  weights: models.ViT_H_14_Weights.IMAGENET1K_SWAG_E2E_V1
+
+vit_l_16:
+  metrics:
+    Acc@1: 79.662  # Larger ViT model with improved accuracy over base models
+    Acc@5: 94.638
+    GFLOPS: 61.55
+    Params: 304.3M
+  model_fn: models.vit_l_16
+  resolution: 224
+  weights: models.ViT_L_16_Weights.IMAGENET1K_V1
+
+vit_l_32:
+  metrics:
+    Acc@1: 76.972  # Larger ViT with larger patch size, trading accuracy for reduced compute
+    Acc@5: 93.07
+    GFLOPS: 15.38
+    Params: 306.5M
+  model_fn: models.vit_l_32
+  resolution: 224
+  weights: models.ViT_L_32_Weights.IMAGENET1K_V1

data/raw/.gitignore

@@ -0,0 +1,4 @@
+# Ignore everything in this directory
+*
+# Except this .gitignore file
+!.gitignore