Interactive agglomeration scheme for paintera
Project description
Paintera Interactive Agglomeration Server
pias
provides a server for interactive updates of a nifty
-based agglomeration model for fragments in EM segmentation (but could probably be used in other scenarios, as well). Clients provide positive (two fragments are in the same segment) and negative (two fragments should not be in the same segment) examples for edges (fragment-pairs). Random forest classifiers are trained on these examples to infer weights for all edges of neighboring fragments in a multi-cut problem (constraints for the enforcement of positive/negative examples provided by the client is modeled via appropriate weights).
Installation
The minimum python version is 3.6
.
Besides the packages specified in setup.py
, you will need to install the following packages that are not installable via pip:
nifty
available on conda through thecpape
channelz5py
available on conda through theconda-forge
channel
You can install these dependencies through conda via the aforementioned channels. Otherwise, follow the build instructions on the linked github repositories.
If you do not install pyzmq
through conda
make sure that libzmq
is installed on your system.
pip install git+https://github.com/saalfeldlab/pias
Server Details
pyzmq
is used for communication between server and client. Data should be sent as big endian. The server can be started with the pias
command that is installed with the python package. See pias --help
for usage details. Once started, the server will start multiple sockets, addressed by extensions of the address_base
parameter:
${address_base}-ping
- ping the server at this address to see if it is alive (REQ/REP
)${address_base}-current-solution
- request current solution (REQ/REP
)${address_base}-set-edge-labels
- set labels for edges: (multiples of)(e1, e2, label)
(REQ/REP
) where label is one of{0, 1}
${address_base}-update-solution
- request update of current solution (REQ/REP
)${address_base}-new-solution
- be notified about updates of the current solution (PUB/SUB
)
NOTE: This scheme probably works (reliably) with ipc://
zmq-addresses.
Socket Details
TBD: What kind of input/output does each socket expect/provide?
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
File details
Details for the file pias-0.1.0.dev0.tar.gz
.
File metadata
- Download URL: pias-0.1.0.dev0.tar.gz
- Upload date:
- Size: 13.1 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/1.13.0 pkginfo/1.5.0.1 requests/2.21.0 setuptools/40.8.0 requests-toolbelt/0.9.1 tqdm/4.28.1 CPython/3.7.2
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 8e65d8245c6d9f90094483aae8aa2966835ed57fffaefd4686db2d9898371a0f |
|
MD5 | bb03d2685c2a40d3727b8436b05502a8 |
|
BLAKE2b-256 | 96f418b189c116adada15e8ab46b24227ad6b3f9938f2b20e76a59a0827cb7fe |