search

WisconsinAIVision / yolact_edge

The first competitive instance segmentation approach that runs on small edge devices at real-time speeds.
MIT License
1.28k stars 273 forks source link

I would like to know the same training method as the YolactEdge paper(ICRA 2021). #166

Closed JunsukLee closed 2 years ago

JunsukLee commented 2 years ago

If I follow the 4 commands below in order, will I get the same result as your paper? I want to train on my computer and get the same results from your paper.

//Trains using the base edge config with a batch size of 8 (the default). //Training resnet101_reducedfc.pth python train.py --config=yolact_edge_config

//Pre-train the image based model //Fine-tuning for Youtubevis dataset python train.py --config=yolact_edge_youtubevis_config

//Train the flow (warping) module python train.py --config=yolact_edge_vid_trainflow_config --resume=./weights/yolact_edge_youtubevis_847_50000.pth

//Fine tune the network jointly python train.py --config=yolact_edge_vid_config --resume=./weights/yolact_edge_vid_trainflow_144_100000.pth

haotian-liu commented 2 years ago

Yes, it is how we train the models :)

hamedmh commented 2 years ago

Hi, I would like to train the yolact_edge model for instance segmentation using this dataset:

https://github.com/by256/emps

The images are in one folder, and the annotation images are in another folder. Each object in the annotation image is labeled with a certain value for all of its pixels. How should I modify this dataset to use it for training of yolact_edge? Thanks! Best regards, Hamed