SegFormer-b0-person-segmentation Open-source Model - Accurately Assign Semantic Category Labels to Image Pixels

Segformer B0 Person Segmentation

Developed by s3nh

A semantic segmentation model based on the Segformer architecture, used to assign semantic category labels to each pixel in an image.

Image Segmentation

Transformers

EnglishOpen Source License:Openrail #Semantic Segmentation #Pixel-level Annotation #Image Parsing

Downloads 3,187

Release Time : 2/6/2023

Model Overview

This model is a computer vision technology capable of assigning labels to each pixel in an image, identifying and understanding objects and scenes within the image, and segmenting the image into parts corresponding to different entities.

Model Features

Dense Prediction

Assigns labels to each pixel in an image, achieving precise semantic segmentation.

Efficient Architecture

Based on the Segformer architecture, combining the advantages of Transformer and CNN to provide efficient image segmentation capabilities.

Easy to Use

Provides clear installation and usage examples for quick integration into projects.

Model Capabilities

Image Semantic Segmentation

Pixel-level Classification

Scene Understanding

Use Cases

Computer Vision

Autonomous Driving

Used to identify key elements such as roads, vehicles, and pedestrians, assisting autonomous driving systems in environmental perception.

Medical Image Analysis

Used to segment organs or lesion areas in medical images, assisting in diagnosis and treatment.

🚀 Image Segmentation Project

This project focuses on image segmentation, a computer vision technique that assigns a label to each pixel in an image, enabling the recognition and partitioning of objects and scenes.

🚀 Quick Start

Semantic segmentation is a powerful computer vision technique. It assigns a label to each pixel in an image, representing the semantic class of the objects or regions. This is different from object detection or instance segmentation, as it provides a more detailed understanding of the image by labeling every pixel.

✨ Features

Semantic Understanding: Assigns a label to each pixel, enabling detailed recognition of objects and scenes in an image.
Dense Prediction: Unlike traditional object detection, it provides a label for every pixel, offering a more comprehensive view of the image.

📦 Installation

model = SegformerForSemanticSegmentation.from_pretrained("/notebooks/segformer_5_epoch",
                                                         num_labels=2, 
                                                         id2label=id2label, 
                                                         label2id=label2id, )

💻 Usage Examples

Basic Usage

from torch import nn
import numpy as np
import matplotlib.pyplot as plt

# Transforms
_transform = A.Compose([
    A.Resize(height = 512, width=512), 
    ToTensorV2(), 
])


trans_image = _transform(image=np.array(image))
outputs = model(trans_image['image'].float().unsqueeze(0))
logits = outputs.logits.cpu()
print(logits.shape)


# First, rescale logits to original image size
upsampled_logits = nn.functional.interpolate(logits,
                size=image.size[::-1], # (height, width)
                mode='bilinear',
                align_corners=False)


seg = upsampled_logits.argmax(dim=1)[0]
color_seg = np.zeros((seg.shape[0], seg.shape[1], 3), dtype=np.uint8) # height, width, 3
palette = np.array([[0, 0, 0],[255, 255, 255]])
for label, color in enumerate(palette):
    color_seg[seg == label, :] = color
# Convert to BGR
color_seg = color_seg[..., ::-1]

📄 License

This project is licensed under the OpenRAIL license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご