You can deploy this model and start earning money today!

kerasformers/
segformer_b4_cityscapes_1024

Name: segformer_b4_cityscapes_1024
Availability: InStock

kerasformers

keras

segformer

semantic-segmentation

nvidia/segformer-b4-finetuned-cityscapes-1024-1024

finetune

other

Views

No views yet

See our collection for all versions of SegFormer.

Run SegFormer with Keras 3: JAX, PyTorch, or TensorFlow

GitHub

zeromodels/segformer_b4_cityscapes_1024

Paper: SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers (arXiv:2105.15203) · HF Papers

SegFormer pairs a hierarchical transformer encoder (MiT) with a lightweight all-MLP decoder. The encoder produces features at four scales, and the decoder upsamples, concatenates, and projects them. Sequence-reduction attention and no positional encoding keep it efficient and resolution-flexible.

For more details on the model, please go to NVIDIA's original model card.

Pure-Keras 3 conversion of nvidia/segformer-b4-finetuned-cityscapes-1024-1024 for zeromodels. One implementation runs unmodified on TensorFlow / Torch / JAX.

This is a semantic segmentation checkpoint (SegFormerSemanticSegment) for Cityscapes (19 classes, MiT-B4, 1024px).

✨ Quick start

python

import os
os.environ["KERAS_BACKEND"] = "torch"  # or "jax" / "tensorflow"

from PIL import Image
from zeromodels.models.segformer import (
    SegFormerSemanticSegment,
    SegFormerImageProcessor,
)

model = SegFormerSemanticSegment.from_weights("zeromodels/segformer_b4_cityscapes_1024")
processor = SegFormerImageProcessor.from_weights("zeromodels/segformer_b4_cityscapes_1024")

image = Image.open("your_image.jpg").convert("RGB")
output = model(processor(image)["pixel_values"], training=False)
result = processor.post_process_semantic_segmentation(
    output, target_size=(image.height, image.width)
)
print(result["unique_classes"], result["class_names"])

Load any SegFormer variant the same way with from_weights("zeromodels/<variant>"):

Variant	Hub	Dataset	Res
`segformer_b0_ade_512`	`zeromodels/segformer_b0_ade_512`	ADE20K	512
`segformer_b1_ade_512`	`zeromodels/segformer_b1_ade_512`	ADE20K	512
`segformer_b2_ade_512`	`zeromodels/segformer_b2_ade_512`	ADE20K	512
`segformer_b3_ade_512`	`zeromodels/segformer_b3_ade_512`	ADE20K	512
`segformer_b4_ade_512`	`zeromodels/segformer_b4_ade_512`	ADE20K	512
`segformer_b5_ade_640`	`zeromodels/segformer_b5_ade_640`	ADE20K	640
`segformer_b0_cityscapes_768`	`zeromodels/segformer_b0_cityscapes_768`	Cityscapes	768
`segformer_b0_cityscapes_1024`	`zeromodels/segformer_b0_cityscapes_1024`	Cityscapes	1024
`segformer_b1_cityscapes_1024`	`zeromodels/segformer_b1_cityscapes_1024`	Cityscapes	1024
`segformer_b2_cityscapes_1024`	`zeromodels/segformer_b2_cityscapes_1024`	Cityscapes	1024
`segformer_b3_cityscapes_1024`	`zeromodels/segformer_b3_cityscapes_1024`	Cityscapes	1024
`segformer_b4_cityscapes_1024`	`zeromodels/segformer_b4_cityscapes_1024`	Cityscapes	1024
`segformer_b5_cityscapes_1024`	`zeromodels/segformer_b5_cityscapes_1024`	Cityscapes	1024

Tips

Set KERAS_BACKEND before importing Keras / zeromodels.
Prefer SegFormerImageProcessor.from_weights(...) so the resize matches the variant (ADE B5 is 640; Cityscapes is often 1024).
See SegFormer docs and Loading Weights.
Community / upstream safetensors still work via the hf: prefix, e.g. SegFormerSemanticSegment.from_weights("hf:nvidia/segformer-b4-finetuned-cityscapes-1024-1024").

Special Thanks

A huge thank you to the NVIDIA SegFormer authors for creating and releasing these models.

License: see the NVIDIA SegFormer LICENSE (Hub tag: other).

kerasformers/segformer_b4_cityscapes_1024