MediaPipe DeepLab-v3 (Image Segmenter)

General-purpose DeepLab-v3 semantic segmentation model (deeplab_v3.tflite, 2.8 MB, FP32, 257 x 257 input) segmenting 21 Pascal VOC classes (background, person, cat, dog, potted plant, vehicles, furniture) for the MediaPipe Image Segmenter (ImageSegmenter) task.

Documentation: MediaPipe Image Segmenter Docs

Samples: Android · iOS · Python Colab · Web · Flutter

Try live: Web Demo

Uses atrous spatial pyramid pooling (ASPP) over [1, 257, 257, 3] RGB input to produce [1, 257, 257, 21] per-pixel logits/probabilities. Embeds labelmap.txt / labels.txt in TFLITE_METADATA:

0: background, 1: aeroplane, 2: bicycle, 3: bird, 4: boat, 5: bottle, 6: bus, 7: car, 8: cat, 9: chair, 10: cow, 11: dining table, 12: dog, 13: horse, 14: motorbike, 15: person, 16: potted plant, 17: sheep, 18: sofa, 19: train, 20: tv.

Files

File Precision Size SHA-256
deeplab_v3.tflite float32 2.78 MB (2,780,176 B) ff36e24d40547fe9e645e2f4e8745d1876d6e38b332d39a82f0bf0f5d1d561b3

Download

pip install -U huggingface_hub
hf download litert-community/MediaPipe-ImageSegmenter-DeepLab-V3 deeplab_v3.tflite --local-dir .

Inputs, Outputs, and Model Specifications

deeplab_v3.tflite

  • Embedded TFLITE_METADATA: name ImageSegmenter · version 1 · author MediaPipe · packed associated files: labelmap.txt, labels.txt
Direction Tensor Shape Type Notes
Input sub_7 [1, 257, 257, 3] float32 image — Input image to be processed. — Normalization: mean=[127.5], std=[127.5]
Output ResizeBilinear_3 [1, 257, 257, 21] float32 segmentation_masks — Masks over the target objects with high accuracy. — Labels: labels.txt

Use with MediaPipe Tasks

Python Quickstart

pip install -U mediapipe huggingface_hub
import mediapipe as mp
from huggingface_hub import hf_hub_download

model_path = hf_hub_download(
    repo_id="litert-community/MediaPipe-ImageSegmenter-DeepLab-V3",
    filename="deeplab_v3.tflite",
)

BaseOptions = mp.tasks.BaseOptions
ImageSegmenter = mp.tasks.vision.ImageSegmenter
ImageSegmenterOptions = mp.tasks.vision.ImageSegmenterOptions
VisionRunningMode = mp.tasks.vision.RunningMode

options = ImageSegmenterOptions(
    base_options=BaseOptions(model_asset_path=model_path),
    running_mode=VisionRunningMode.IMAGE,
    output_category_mask=True,
    output_confidence_masks=False,
)

with ImageSegmenter.create_from_options(options) as runner:
    mp_image = mp.Image.create_from_file("input.jpg")
    result = runner.segment(mp_image)
    print("Category mask shape:", result.category_mask.numpy_view().shape)

Cross-Platform Guides & Samples

Latency Benchmarks

End-to-end MediaPipe Task pipeline average latency reported by the Image Segmenter documentation:

Model Variant Device CPU Latency GPU Latency
DeepLab-V3 Pixel 6 123.93 ms 103.30 ms

Intended Use, Limitations, and Fairness

  • Intended applications: General-purpose scene segmentation across the 20 foreground Pascal VOC object classes plus background (photo effects, foreground extraction, scene understanding).
  • Limitations & considerations: Restricted to the 21 Pascal VOC categories at 257 x 257 model resolution; fine boundaries or out-of-vocabulary objects are not separated. No standalone Google fairness PDF is published for DeepLab-v3 on the MediaPipe documentation page.

Provenance

Every model binary in this repository is byte-identical (unmodified FlatBuffer weights and embedded TFLITE_METADATA bytes) to the official artifact served by Google from storage.googleapis.com/mediapipe-models.

Repo File Upstream URL GCS Generation Upstream Last-Modified Pinned Bucket Version
deeplab_v3.tflite deeplab_v3.tflite 1682480021097695 Wed, 26 Apr 2023 03:33:41 GMT /1/

License and Citation

  • License: apache-2.0 — Distributed under Apache License, Version 2.0 alongside the MediaPipe release artifacts.
Downloads last month
-
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Collection including litert-community/MediaPipe-ImageSegmenter-DeepLab-V3