Mask2former Deployment

🚀 Mask2Former for Semantic Segmentation

This repository offers a fine - tuned Mask2Former model for semantic segmentation tasks. It can predict segmentation masks on input images, leveraging the facebook/mask2former-swin-large-cityscapes-semantic pre - trained model.

🚀 Quick Start

This section will guide you through using the Mask2Former model for semantic segmentation.

✨ Features

• General - Purpose Framework: Mask2Former serves as a versatile solution for various mask prediction tasks, including semantic, instance, and panoptic segmentation.
• Fine - Tuned for Semantic Segmentation: This specific version has been optimized for semantic segmentation, making it suitable for applications like road scene understanding and autonomous driving.

📦 Installation

First, make sure you have the necessary libraries installed:

pip install transformers torch torchvision pillow matplotlib

💻 Usage Examples

Basic Usage

You can utilize this model with the transformers library from Hugging Face. The following example demonstrates how to load the model, process an image, and visualize the output.

from transformers import AutoImageProcessor, Mask2FormerForUniversalSegmentation
from PIL import Image
import torch
import matplotlib.pyplot as plt

# Load the processor and model
model_name = "saninmohammedn/mask2former-deployment"
processor = AutoImageProcessor.from_pretrained(model_name)
model = Mask2FormerForUniversalSegmentation.from_pretrained(model_name)

# Load an input image
image_path = "your_image.jpg"  # Replace with your image path
image = Image.open(image_path).convert("RGB")

# Prepare the image for the model
inputs = processor(images=image, return_tensors="pt")

# Perform inference
with torch.no_grad():
    outputs = model(**inputs)

# Post-process the predicted segmentation map
predicted_map = processor.post_process_semantic_segmentation(
    outputs, target_sizes=[image.size[::-1]]
)[0].cpu().numpy()

# Visualize the input and predicted segmentation map
plt.figure(figsize=(10, 5))

# Display original image
plt.subplot(1, 2, 1)
plt.imshow(image)
plt.title("Original Image")
plt.axis("off")

# Display predicted segmentation map
plt.subplot(1, 2, 2)
plt.imshow(predicted_map, cmap="jet")
plt.title("Predicted Segmentation Map")
plt.axis("off")

plt.tight_layout()
plt.show()

📄 License

The license for this project is under "other".