【PaddleClas】 Common commands

1. Generate data set

Need modification build_product.yaml

python python/build_gallery.py -c configs/build_product.yaml -o IndexProcess.data_file="D:\Paddle\PaddleClas\out\recognition_demo_data_v1.0\gallery_sg\data_file.txt" -o IndexProcess.index_path="D:\Paddle\PaddleClas\out\recognition_demo_data_v1.0\gallery_sg\index"

3. forecast

python tools/infer.py -c configs/ppyolo/ppyolo_r50vd_dcn_voc.yml --infer_dir=/home/aiuser/mtl/data/sg/test --output_dir=./sg_out_put/ --draw_threshold=0.5  -o weights=./output/ppyolo_r50vd_dcn_voc/model_final

4. export

python tools/export_model.py -c configs/ppyolo/ppyolo_r50vd_dcn_voc.yml \
        --output_dir=./inference_model \
        -o weights=output/ppyolo_r50vd_dcn_voc/model_final

Optional parameter list

The following list can be accessed through --help see

FLAG	Support script	purpose	The default value is	remarks
-c	ALL	Specify profile	None	Please refer to Configuration module
-o	ALL	Set the parameters in the configuration file	None	Use -o Configuration compared to -c The selected profile has a higher priority . for example ：-o use_gpu=False max_iter=10000
-r/–resume_checkpoint	train	Resume training from a checkpoint	None	-r output/faster_rcnn_r50_1x/10000
–eval	train	Whether to test while training	False
–output_eval	train/eval	Edit the evaluation and save json route	current path	--output_eval ./json_result
–fp16	train	Whether to use mixed accuracy training mode	False	Need to use GPU Training
–loss_scale	train	Set the scaling of the loss value in the mixed accuracy training mode	8.0	It needs to be opened first --fp16 After use
–json_eval	eval	Whether to pass the existing bbox.json perhaps mask.json To assess the	False	json The file path is --output_eval Set in
–output_dir	infer	Output post prediction visualization file	./output	--output_dir output
–draw_threshold	infer	Visual time score threshold	0.5	--draw_threshold 0.7
–infer_dir	infer	Picture folder path for prediction	None
–infer_img	infer	Picture path for prediction	None	Compare with --infer_dir Higher priority
–use_vdl	train/infer	Whether to use VisualDL Record data , And then in VisualDL The panel shows	False	VisualDL Need to be Python>=3.5
–vdl_log_dir	train/infer	Appoint VisualDL Storage path for recording data	train:vdl_log_dir/scalar infer: vdl_log_dir/image	VisualDL Need to be Python>=3.5