Junjue-Wang / EarthVQA

[AAAI 2024] EarthVQA: Towards Queryable Earth via Relational Reasoning-Based Remote Sensing Visual Question Answering
80 stars 2 forks source link

Issue to train the segmentation network by own #15

Open Liuyan9627 opened 3 months ago

Liuyan9627 commented 3 months ago

Thanks for your awesome work. I want to train a segmentation network on the provided dataset, via run:

# 1 train a segmentation model
sh ./scripts/train_sfpnr50.sh
# 2 generate segmentation features and pse-masks
sh ./scripts/generate_segfeats.sh
# 3 train SOBA
sh ./scripts/train_soba.sh

when I run the: sh ./scripts/train_sfpnr50.sh an error occurs:

usage: launch.py [-h] [--nnodes NNODES] [--nproc_per_node NPROC_PER_NODE] [--rdzv_backend RDZV_BACKEND] [--rdzv_endpoint RDZV_ENDPOINT] [--rdzv_id RDZV_ID] [--rdzv_conf RDZV_CONF] [--standalone]
                 [--max_restarts MAX_RESTARTS] [--monitor_interval MONITOR_INTERVAL] [--start_method {spawn,fork,forkserver}] [--role ROLE] [-m] [--no_python] [--run_path] [--log_dir LOG_DIR]
                 [-r REDIRECTS] [-t TEE] [--node_rank NODE_RANK] [--master_addr MASTER_ADDR] [--master_port MASTER_PORT] [--use_env]
                 training_script ...
launch.py: error: argument --master_port: invalid int value: 'train_lovedav2_seg.py'

Could you please kindly remind me how to fix it? Thanks.

Liuyan9627 commented 3 months ago

This problem is solved: I modify the --master_port $RANDOM as -- master_port=1234. Then under the data folder: I change the name of file 'distribute' to 'distribute.py' the training code can run.