semiweeds's Introduction

Performance Evaluation of Semi-supervised Learning Frameworks for Multi-Class Weed Detection

More instructions coming soon...

Installation

Follow the installation instructions of INSTALL.md

Training and testing commands

Train supervised baseline. For example, using 20% of labeled data

python train_net.py \
       --num-gpus 2 \
       --config configs/FCOS/fcos_R_50_ut2_sup20_run0.yaml \
        SOLVER.IMG_PER_BATCH_LABEL 4 SOLVER.IMG_PER_BATCH_UNLABEL 4 SOLVER.IMS_PER_BATCH 4 \
        OUTPUT_DIR ./xx/ \
        TEST.EVAL_PERIOD 2000 \
        SEED 1 \
        SEMISUPNET.Trainer baseline

Train semi-supervised. For example, using 20% of labeled data + 80% unlabeled data

python train_net.py \
      --num-gpus 2 \
      --config configs/FCOS/fcos_R_50_ut2_sup20_run0.yaml \
       SOLVER.IMG_PER_BATCH_LABEL 4 SOLVER.IMG_PER_BATCH_UNLABEL 4 SOLVER.IMS_PER_BATCH 4 \
       OUTPUT_DIR ./xx/ \
       TEST.EVAL_PERIOD 2000 \
       SEED 1

Train semi-supervised. For example, using 20% of labeled data + 80% unlabeled data

python train_net.py \
        --eval-only \
	--num-gpus 2 \
	--config configs/FCOS/fcos_R_50_ut2_sup20_run0.yaml \
	SOLVER.IMG_PER_BATCH_LABEL 4 SOLVER.IMG_PER_BATCH_UNLABEL 4 SOLVER.IMS_PER_BATCH 4 \
	MODEL.WEIGHTS ./model_final.pth \
	OUTPUT_DIR ./xx/ \
        SEMISUPNET.Trainer baseline \
	SEED 1