Skip to content
Other

DigitNet v2 Display Band

ONNX model for reading seven-segment displays

2 min read

This tool recognizes digits on seven-segment liquid crystal screens from local images (JPEG or PNG) using specified block coordinates. The model is based on DigitNet v2 weights by Chakib Belgaid, adapted to run on a CPU in ONNX format (while adding background evaluation for the entire display band and preserving core external readings). It accepts character crops as input, outputting probabilities for numbers from 0 to 9 or a non-digit flag at a confidence threshold of 0.60. This solution benefits developers of smart meters, IoT systems, and offline autonomous applications for automated meter reading.

Key features

  • Lightweight architecture: contains only 10,123 parameters with an ONNX file size of 42,410 bytes.
  • Offline operation: completely autonomous execution on a CPU after a one-time download of dependencies and the archive.
  • Hybrid approach: allows re-evaluating bounding boxes rejected by the primary classifier.
  • Convenient distribution: the archive includes launch scripts, sample images, and JSON annotations.

What you can do

  • Utility automation: integrate the model into a local image processing pipeline for electricity or water meters.
  • Fine-tuning and testing: evaluate recognition accuracy on custom datasets with digit coordinates.
  • IoT integration: use the lightweight ONNX file in custom Python applications for edge devices.

Example

curl -fL '[https://modelscope.cn/models/nano0412/inchwork-digitnet-v2-display-band-onnx/resolve/v0.1.0/digitnet-display-band-cpu-v0.1.0.zip](https://modelscope.cn/models/nano0412/inchwork-digitnet-v2-display-band-onnx/resolve/v0.1.0/digitnet-display-band-cpu-v0.1.0.zip)' -o digitnet.zip
python3.12 -m zipfile -e digitnet.zip .
cd E-009-P-digitnet-cpu
python3.12 -m venv .venv
.venv/bin/python -m pip install -r requirements.txt
.venv/bin/python digitnet_pilot.py examples/display.png examples/boxes.json
Read moreDetails
• Architecture: uint8 tensor input [N,1,32,24], where 0 is background and 255 represents the deepest strokes.
• Output: probabilities [N,11] with a class activation threshold ≥0.60.
• Benchmarks: on a test set of 904 bounding boxes, the base DigitNet achieved 737 correct answers and 166 rejections (increasing up to 790 correct answers when combined with a fallback method).
• Requirements: Python 3.12, runs on a standard CPU with no GPU required.
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.0 GB · ready files0.5 GB
  • FP8 / INT8recommendedweights 0.0 GB0.5 GB
  • 4-bitweights 0.0 GB0.5 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · BF16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # kang0110/inchwork-digitnet-v2-display-band-onnx
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download kang0110/inchwork-digitnet-v2-display-band-onnx --local-dir ~/models/inchwork-digitnet-v2-display-band-onnx
    # The model is in ~/models. How to run it is in the model card (link above).
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of inchwork-digitnet-v2-display-band-onnx (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

kang0110/inchwork-digitnet-v2-display-band-onnx
Author
kang0110
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments

Other

K-ALUAN

K-ALUAN