Module for input, preprocessing, and deployment of the CV and SSC based models
This module was based on the Artificial Neural Networks developed by Luisa Vieira Lucchese, available at https://github.com/luisalucchese/ann-ssc-module
- datagen goes through reaches and finds tiles for each of the lines within date range
- stores them in a input json
- process_ssc reads in this and uses the array index to pick out a tile
- checks tile for overlap with nodes
- predicts on nodes in tile
- writes to tile level efs
- ssc_combine module to coalate the results
sudo docker run -v /mnt/flpe/consensus:/data/consensus -v /mnt/input:/data/input -v /home/ec2-user/:/root/ ssc_deploy --consensus_dir /data/consensus -j /data/input/global_HLS_filtered_plus_minus_1_day_compressed_no_dupes -o /data/input/ssc_july_16/results -d /data/input --ckpt_path /data/input/ssc/models/model_static_files/rangel_checks/multitask_ckpts/deeplabv3p_distrib.pth.tar --ann_model_dir /data/input/ssc/models/model_static_files/final_model_static_files -i 2 -c > output.log 2>&1