Skip to main content

Browse the archive

Revision 0535d452dd426ba62d6d312766906ddd1230c7eb authored by Andrey Zhavoronkov on 02 November 2023, 15:59:39 UTC, committed by GitHub on 02 November 2023, 15:59:39 UTC 
Chunk preparation optimization (#7081)
 
This PR speeds up the preparation of chunks by: 
1. loading images once instead of twice in each writer,
2. as well as by allowing simultaneous preparation of more than 1 chunk
using multithreading.
This allows to reduce the time for preparation of chunks for 4895 images
from 0:04:36 to 0:01:20 in case of preparation of 3 chunks in parallel
and 0:02:46 in case of 1 chunk in my environment.

Co-authored-by: Maria Khrustaleva <maya17grd@gmail.com>
1 parent 1f8d5d3
Raw File
config.ts 
// Copyright (C) 2019-2022 Intel Corporation
// Copyright (C) 2022 CVAT.ai Corporation
//
// SPDX-License-Identifier: MIT

const config = {
    backendAPI: '/api',
    organization: {
        organizationID: null,
        organizationSlug: null,
    },
    origin: '',
    uploadChunkSize: 100,
    removeUnderlyingMaskPixels: false,
    onOrganizationChange: null,
};

export default config;
back to top