Skip to content

UnicornGlade/ObjectsDetection

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

32 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

How to install mmcv on Python 3.12 (currently doesn't work)

mmcv doesn't officially support Python 3.12+, so it is recommended to use Python 3.11 - see open-mmlab/mmcv#3263

But it seems that it should work on Python 3.12.10 win64 - open-mmlab/mmcv#3263 (comment)

Installation

Install torch with CUDA support:

pip install torch==2.3.0+cu121 torchvision==0.18.0+cu121 --extra-index-url https://download.pytorch.org/whl/cu121

Install other dependencies:

# It will:
# - install openmim
# - install mmcv, mmdet
# - if Python 3.11 - download pre-trained weights rtmdet_tiny_8xb32-300e_coco_20220902_112414-78e30dcc.pth
pip install -e .

Download dataset with partial labeling (LabelImg was used) and unzip it files into data/dataset01

Download pretrained model weights and put it into pretrained directory. On Python 3.11 these weights were downloaded automatically via calling mim download mmdet --config rtmdet_tiny_8xb32-300e_coco --dest pretrained (but this requires pip install openmim, and it fails on Python 3.12, while works on Python 3.11).

Usage

# generate dataset from manual labels
python src/create_tiles_dataset.py --input_dir data/dataset01 --output_dir data/dataset01_tiles640

# run training on data/dataset01_tiles640 with frozen backbone (i.e. fine-tune)
python src/train_rtmdet.py

python src/export_onnx.py --img data/dataset01/P1000101_1400494888594.JPG --cfg src/configs/rtmdet_car.py --ckpt models/ --out models/exported_model.onnx

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages