Skip to content
ModelListedExternal
Download

Segformer Environment 1

This repository contains a fine-tuned SegFormer model for semantic segmentation of outdoor scenes, specifically targeting:

Sourcehuggingface.co/Dinusharg/segformer_environment_1/tree/50a432a96fc13267dfb649082c1a54a24c18c0ac
KindOtherVersionRevision 50a432a

Model source

Kind
Other
File size
14.2 MB
Version
vRevision 50a432a
Updated
Sep 15, 2026
Downloads
10
Source
Hugging Face

Source description

SegFormer Semantic Segmentation (Road / Grass / Footpath / Water)

This repository contains a fine-tuned SegFormer model for semantic segmentation of outdoor scenes, specifically targeting:

  • background
  • footpath
  • grass
  • road
  • water

The model is designed for applications such as:

  • Autonomous navigation
  • Robotics perception
  • Scene understanding
  • Smart city / mapping solutions

Model Details

  • Architecture: SegFormer
  • Framework: Hugging Face Transformers
  • Input Size: 224 × 224
  • Task: Semantic Segmentation
  • Number of Classes: 5

Class Labels

IDLabel
0background
1footpath
2grass
3road
4water

Quick Start

1. Install Dependencies

pip install transformers torch pillow

2. Run Inference

from transformers import SegformerForSemanticSegmentation, SegformerImageProcessor
from PIL import Image
import torch
import torch.nn.functional as F
import numpy as np

MODEL_ID = "Dinusharg/segformer_environment_1"
IMAGE_PATH = "input_image_path"

OUT_MASK = "segmented_mask.png"
OUT_OVERLAY = "segmented_overlay.png"

palette = {
    0: (0, 0, 0),        # background
    1: (255, 0, 0),      # footpath
    2: (0, 255, 0),      # grass
    3: (128, 128, 128),  # road
    4: (0, 0, 255),      # water
}

processor = SegformerImageProcessor.from_pretrained(MODEL_ID)
model = SegformerForSemanticSegmentation.from_pretrained(MODEL_ID)

image = Image.open(IMAGE_PATH).convert("RGB")
inputs = processor(images=image, return_tensors="pt")

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

upsampled_logits = F.interpolate(
    outputs.logits,
    size=image.size[::-1],
    mode="bilinear",
    align_corners=False
)

pred = upsampled_logits.argmax(dim=1)[0].cpu().numpy()

print("Unique predicted classes:", sorted(set(pred.flatten().tolist())))
print("Labels:", model.config.id2label)


color_mask = np.zeros((pred.shape[0], pred.shape[1], 3), dtype=np.uint8)

for class_id, color in palette.items():
    color_mask[pred == class_id] = color

Image.fromarray(color_mask).save(OUT_MASK)
print(f"Saved mask: {OUT_MASK}")


ALPHA = 0.5

image_np = np.array(image).astype(np.float32)
mask_np = color_mask.astype(np.float32)

overlay = (image_np * (1 - ALPHA) + mask_np * ALPHA).clip(0, 255).astype(np.uint8)

Image.fromarray(overlay).save(OUT_OVERLAY)
print(f"Saved overlay: {OUT_OVERLAY}")

This will returns the colored masked and overlay masked images.

Hugging Face Space (Live Demo)

Try the model interactively:

Launch Demo Space

Upload an image and visualize segmentation results instantly.

Sample Results

Input ImageSegmentation MaskOverlay Output
imageimageimage
imageimageimage
imageimageimage

Contact

For questions or collaboration:

Sources

1 source
Hugging FacePrimaryVersion 50a432a96fc13267dfb649082c1a54a24c18c0ac
huggingface.co/Dinusharg/segformer_environment_1/tree/50a432a96fc13267dfb649082c1a54a24c18c0ac

Verified Sep 15

Model artifacts

1 artifact

model.safetensors

safetensors · 14.2 MB · SHA-256 0191def7ebe9…00c7 · Hugging Face

Download

Source excerpts

2 excerpts
ReadmeSep 15huggingface.co/Dinusharg/segformer_environment_1

--- license: agpl-3.0 library_name: transformers pipeline_tag: image-segmentation tags: - segformer - semantic-segmentation - computer-vision - environment - footpath - grass - road - water - robotic - autonomous navigation - Scene understanding --- # SegFormer Semantic Segmentation (Road / Grass / Footpath / Water) This repository contains a fine-tuned **SegFormer** model for semantic segmentation of outdoor scenes, specifically targeting: - background - footpath - grass - road - water The model is designed for applications such as: - Autonomous navigation - Robotics perception - Scene understanding - Smart city / mapping solutions ## Model Details - **Architecture**: SegFormer - **Framework**: Hugging Face Transformers - **Input Size**: 224 × 224 - **Task**: Semantic Segmentation - **Number of Classes**: 5 ## Class Labels |ID |Label | |---|----------| |0 |background| |1 |footpath | |2 |grass | |3 |road | |4 |water | ## Quick Start ### 1. Install Dependencies ``` pip install transformers torch pillow ``` ### 2. Run Inference ```python from transformers import SegformerForSemanticSegmentation, SegformerImageProcessor from PIL import Image import torch import torch.nn.functional as F import numpy as np MODEL_ID = "Dinusharg/segformer_environment_1" IMAGE_PATH = "input_image_path" OUT_MASK = "segmented_mask.png" OUT_OVERLAY = "segmented_overlay.png" palette = { 0: (0, 0, 0), # background 1: (255, 0, 0), # footpath 2: (0, 255, 0), # grass 3: (128, 128, 128), # road 4: (0, 0, 255), # water } processor = SegformerImageProcessor.from_pretrained(MODEL_ID) model = SegformerForSemanticSegmentation.from_pretrained(MODEL_ID) image = Image.open(IMAGE_PATH).convert("RGB") inputs = processor(images=image, return_tensors="pt") with torch.no_grad(): outputs = model(**inputs) upsampled_logits = F.interpolate( outputs.logits, size=image.size[::-1], mode="bilinear", align_corners=False ) pred = upsampled_logits.argmax(dim=1)[0].cpu().numpy() print("Unique predicted classes:", sorted(set(pred.flatten().tolist()))) print("Labels:", model.config.id2label) color_mask = np.zeros((pred.shape[0], pred.shape[1], 3), dtype=np.uint8) for class_id, color in palette.items(): color_mask[pred == class_id] = color Image.fromarray(color_mask).save(OUT_MASK) print(f"Saved mask: {OUT_MASK}") ALPHA = 0.5 image_np = np.array(image).astype(np.float32) mask_n...

NoteSep 15huggingface.co/Dinusharg/segformer_environment_1

Source context: 10 downloads · 0 likes · Pipeline image-segmentation · Library transformers · Repo Dinusharg/segformer_environment_1