Segformer B0 512x1024 City 160k

🚀 Segformer Model Card

A model for image segmentation using the Segformer architecture in PyTorch

🚀 Quick Start

You can quickly start using the Segformer model by following these steps:

1. Install the necessary libraries.
2. Run inference on an image.

Click the button below to open the example in Google Colab:

Installation

pip install -U segmentation_models_pytorch albumentations

Usage Examples

Basic Usage

import torch
import requests
import numpy as np
import albumentations as A
import segmentation_models_pytorch as smp

from PIL import Image

device = "cuda" if torch.cuda.is_available() else "cpu"

# Load pretrained model and preprocessing function
checkpoint = "smp-hub/segformer-b0-512x1024-city-160k"
model = smp.from_pretrained(checkpoint).eval().to(device)
preprocessing = A.Compose.from_pretrained(checkpoint)

# Load image
url = "https://huggingface.co/datasets/hf-internal-testing/fixtures_ade20k/resolve/main/ADE_val_00000001.jpg"
image = Image.open(requests.get(url, stream=True).raw)

# Preprocess image
np_image = np.array(image)
normalized_image = preprocessing(image=np_image)["image"]
input_tensor = torch.as_tensor(normalized_image)
input_tensor = input_tensor.permute(2, 0, 1).unsqueeze(0)  # HWC -> BCHW
input_tensor = input_tensor.to(device)

# Perform inference
with torch.no_grad():
    output_mask = model(input_tensor)

# Postprocess mask
mask = torch.nn.functional.interpolate(
    output_mask, size=(image.height, image.width), mode="bilinear", align_corners=False
)
mask = mask.argmax(1).cpu().numpy()  # argmax over predicted classes (channels dim)

✨ Features

The Segformer model has the following features:

• Pretrained models: You can easily load a pretrained model from the Hub.
• Flexible initialization: You can customize the model initialization parameters.

📦 Installation

To install the necessary libraries, run the following command:

pip install -U segmentation_models_pytorch albumentations

💻 Usage Examples

Basic Usage

The above example shows how to load a pretrained model and perform inference on an image.

Advanced Usage

You can customize the model initialization parameters as follows:

model_init_params = {
    "encoder_name": "mit_b0",
    "encoder_depth": 5,
    "encoder_weights": None,
    "decoder_segmentation_channels": 256,
    "in_channels": 3,
    "classes": 19,
    "activation": None,
    "aux_params": None
}

📚 Documentation

• Library: https://github.com/qubvel/segmentation_models.pytorch
• Docs: https://smp.readthedocs.io/en/latest/

📄 License

This model is licensed under the terms specified in the following link: https://github.com/NVlabs/SegFormer/blob/master/LICENSE

Dataset

The model is trained on the Cityscapes dataset.

This model has been pushed to the Hub using the PytorchModelHubMixin