Segformer-b0 Open-source Semantic Segmentation Model - Free Deployment, Specially Designed for Sidewalk Scene Optimization

Home

Segformer B0 Finetuned Segments Sidewalk 4

Developed by jakka

Lightweight semantic segmentation model based on MIT-B0 architecture, optimized for sidewalk scenarios

Image Segmentation

Transformers

Open Source License:Apache-2.0 #Sidewalk Semantic Segmentation #Lightweight Model #Urban Street Scene Analysis

Downloads 15

Release Time : 5/30/2022

Model Overview

This model is a fine-tuned version of SegFormer-B0 on the segments/sidewalk-semantic dataset, suitable for semantic segmentation tasks in urban sidewalk scenarios, capable of identifying various pavement elements and obstacles.

Model Features

Lightweight and Efficient

Based on the B0 small architecture, suitable for deployment in resource-constrained environments

Optimized for Sidewalk Scenarios

Specially fine-tuned for urban sidewalk semantic segmentation tasks

Multi-category Recognition

Supports recognition of 30+ pavement element categories (including invalid categories)

Model Capabilities

Image Semantic Segmentation

Pavement Element Recognition

Obstacle Detection

Use Cases

Smart Cities

Sidewalk Accessibility Inspection

Automatically identifies obstacles and pavement defects that affect passage

Achieved 93.3% accuracy in identifying flat pavement on the validation set

Autonomous Driving

Pedestrian Area Delineation

Provides autonomous driving systems with identifiable passable areas

Achieved 78.4% IoU for sidewalks

🚀 segformer-b0-finetuned-segments-sidewalk-4

This model is a fine - tuned version of nvidia/mit-b0 on the segments/sidewalk - semantic dataset. It offers valuable results for image segmentation tasks, specifically on the evaluation set.

🚀 Quick Start

This section provides a high - level overview of the model and its performance. The model achieves the following results on the evaluation set:

Loss: 1.6258
Mean Iou: 0.1481
Mean Accuracy: 0.1991
Overall Accuracy: 0.7316
Per Category Iou: [nan, 0.4971884694242825, 0.7844619900838784, 0.0, 0.10165655377640956, 0.007428563507709108, nan, 4.566798099115959e - 06, 0.0, 0.0, 0.5570746278221521, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.534278997386317, 0.0, 0.0, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.7557693923373933, 0.5270379031768208, 0.8254522211471568, 0.0, 0.0, 0.0, 0.0]
Per Category Accuracy: [nan, 0.8698779680369205, 0.9122325676343133, 0.0, 0.10179229832932858, 0.007508413919135004, nan, 4.566798099115959e - 06, 0.0, 0.0, 0.8968168359562617, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.8492049383357001, 0.0, 0.0, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.9388033874781816, 0.6627890453030717, 0.9334458854084583, 0.0, 0.0, 0.0, 0.0]

📚 Documentation

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

🔧 Technical Details

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 6e - 05
train_batch_size: 32
eval_batch_size: 32
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
lr_scheduler_type: linear
num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Per Category Iou	Per Category Accuracy
1.7912	1.0	25	1.6392	0.1412	0.1911	0.7210	[nan, 0.48942576059104514, 0.7754689525048201, 0.0, 0.031932013148008094, 0.004348266117522573, nan, 1.5527099355168697e - 05, 0.0, 0.0, 0.5356571432088642, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.5243044552616699, 0.0, 0.0, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.7355207837531991, 0.4479559177066271, 0.8315839315332364, 0.0, 0.0, 0.0, 0.0]	[nan, 0.8476069713517648, 0.9129050708992534, 0.0, 0.03194435645315849, 0.004370669306327572, nan, 1.552711353699426e - 05, 0.0, 0.0, 0.897824434787493, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.8555478632753987, 0.0, 0.0, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.9510113270409175, 0.5116786406550935, 0.9122706949370997, 0.0, 0.0, 0.0, 0.0]
1.7531	2.0	50	1.6258	0.1481	0.1991	0.7316	[nan, 0.4971884694242825, 0.7844619900838784, 0.0, 0.10165655377640956, 0.007428563507709108, nan, 4.566798099115959e - 06, 0.0, 0.0, 0.5570746278221521, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.534278997386317, 0.0, 0.0, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.7557693923373933, 0.5270379031768208, 0.8254522211471568, 0.0, 0.0, 0.0, 0.0]	[nan, 0.8698779680369205, 0.9122325676343133, 0.0, 0.10179229832932858, 0.007508413919135004, nan, 4.566798099115959e - 06, 0.0, 0.0, 0.8968168359562617, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.8492049383357001, 0.0, 0.0, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.9388033874781816, 0.6627890453030717, 0.9334458854084583, 0.0, 0.0, 0.0, 0.0]

Framework versions

Transformers 4.19.2
Pytorch 1.11.0
Datasets 2.2.2
Tokenizers 0.12.1

📄 License

This model is licensed under the Apache - 2.0 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