Problem with pseudo-label generation

[Rzy1221]

Dear author, when I run this command，Create folder root/data/pseudo_labels/5shots/ and generate pseudo labels by running:

CUDA_VISIBLE_DEVICES=0 python tools/train_mt_part.py --cfg experiments/ap10k/hrnet/w32_256x192_adam_lr1e-3.yaml --animalpose --generate_pseudol --pretrained output/output_animal_hrnet5_part/model_best.pth OUTPUT_DIR test MODEL.NAME pose_hrnet_part TRAIN.BATCH_SIZE_PER_GPU 32 GPUS [0,] LABEL_PER_CLASS 5

There's been an error. My CPU usage is high and seems to be stuck

My first step of training with 5 images per species went well, but I got stuck generating false labels. Looking forward to your answer.

[chaneyddtt]

Hi @Rzy1221, it is not stuck, it is generating pseudo labels and the process is quite slow.

[Rzy1221]

Thanks for your reply！ I got the result in a few longer hours.