edolview 0.1.2

A remote tool for edolview

edolview

Lightweight Python client for sending images (numpy or torch tensors) to the edolview-rs application for interactive inspection.

Features

• Accepts numpy.ndarray or torch.Tensor input
• Shape normalization: (B,C,H,W), (C,H,W), (H,W), (H,W,C) -> (H,W,C)
• Channels 1–4 supported
• Optional area downscale (fast average pooling)
• Optional compression: PNG for integer types (if OpenCV available) or zlib for float data
• Optional float16 conversion for reduced bandwidth

Install

pip install edolview

Usage

import numpy as np
from edolview import send_image

# Example image (H,W,C) uint8
img = np.random.randint(0, 255, (512, 512, 3), dtype=np.uint8)

# Send to viewer running at host:port
send_image("127.0.0.1", 4567, "random", img, float_to_half=True, do_compression=True)

Or use the class directly:

from edolview import EdolView
viewer = EdolView("127.0.0.1", 4567)
viewer.send_image("frame0", img, do_compression=True)

Torch tensor example

import torch
from edolview import send_image

x = torch.rand(1, 3, 256, 256)  # BCHW float32
send_image("127.0.0.1", 4567, "torch_example", x, do_compression=True)

Downscale

send_image("127.0.0.1", 4567, "big_image", big_img, downscale_factor=4)

API

send_image(host, port, name, image, float_to_half=True, do_compression=False, downscale_factor=1)

EdolView(host, port).send_image(...)

License

MIT. See LICENSE.

Development

# from python/ directory
python -m pip install -e .[dev]
pytest

Publish (manual)

# Ensure version bumped in pyproject.toml
python -m pip install --upgrade build twine
python -m build
python -m twine upload dist/*