https://github.com/kprokofi/simplemaskrcnn

Simple MaskRCNN Instance Segmentation pipeline with training support on XPU (using Intel Extension For Pytorch), GPU and CPU
https://github.com/kprokofi/simplemaskrcnn

Last synced: 17 days ago
JSON representation

Simple MaskRCNN Instance Segmentation pipeline with training support on XPU (using Intel Extension For Pytorch), GPU and CPU

• Host: GitHub
• URL: https://github.com/kprokofi/simplemaskrcnn
• Owner: kprokofi
• License: apache-2.0
• Created: 2024-02-22T12:42:46.000Z (11 months ago)
• Default Branch: main
• Last Pushed: 2024-04-08T15:42:30.000Z (9 months ago)
• Last Synced: 2024-04-08T19:08:06.520Z (9 months ago)
• Language: Python
• Size: 33.2 KB
• Stars: 0
• Watchers: 3
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# SimpleMaskRCNN

Simple MaskRCNN Instance Segmentation pipeline with training support on XPU (using Intel Extension For Pytorch), GPU and CPU

# Installation

1. Install torch either for cuda or for XPU (IPEX)

    ``

    pip install torch==2.1.2 torchvision==0.16.2 torchaudio==2.1.2 --index-url https://download.pytorch.org/whl/cu118

    ``

2. Install requirements

    `` pip install -r requirements.txt ``

# Dataset format

[Datumaro COCO format](https://openvinotoolkit.github.io/datumaro/latest/docs/data-formats/formats/coco.html)

# How to run

```

usage: train.py [-h] [--epochs EPOCHS] --data-root DATA_ROOT [--device DEVICE] [--print-freq PRINT_FREQ] [--image-size IMAGE_SIZE [IMAGE_SIZE ...]] [--batch-size BATCH_SIZE] [--num-workers NUM_WORKERS] [--lr LR] [--wd WD] [--step-lr STEP_LR] [--warmup WARMUP]

options:

  -h, --help            show this help message and exit

  --epochs EPOCHS       number of total epochs to run

  --data-root DATA_ROOT

                        path to dataset

  --device DEVICE       device to use for training

  --print-freq PRINT_FREQ

                        print frequency

  --image-size IMAGE_SIZE [IMAGE_SIZE ...]

                        input image size

  --batch-size BATCH_SIZE

                        batch size

  --num-workers NUM_WORKERS

                        number of workers

  --lr LR               learning rate

  --wd WD               weight decay

  --step-lr STEP_LR     step learning rate. If -1 passed the step lr will be difined automatically as the 0.7 of the epochs. if passed 0 no step lr will be used

  --warmup WARMUP       use warmup learning rate scheduler

  ```