le layer2 -le layer3 # Which layers to extract features from. Patch_core # We now pass all PatchCore-related parameters. log_group IM224_WR50_L2-3_P01_D1024-1024_PS-3_AN-1_S0 -log_project MVTecAD_Results results # Logging details: Name of the run & Name of the overall project folder. log_online # If set, logs results to a Weights & Biases account. save_patchcore_model # If set, saves the patchcore model(s). gpu -seed # Set GPU-id & reproducibility seed. Make sure that it follows the following data tree:
#Amazon anomaly 2 download
To set up the main MVTec AD benchmark, download it from here. However using significantly large input images will incur higher memory cost. In general, the majority of experiments should not exceed 11GB of GPU memory Our results were computed using Python 3.8, with packages and respective version noted in To use them (and replicate training),Ĭheck out sample_evaluation.sh and sample_training.sh.
Python bin/load_and_evaluate_patchcore.py -gpu 0 -seed 0 $savefolder \ĭataset -resize 366 -imagesize 320 " " mvtec $datapathĪ set of pretrained PatchCores are hosted here: add link. Model_flags=( $(for dataset in " " do echo '-p ' $loadpath '/ ' $modelfolder '/models/mvtec_ ' $dataset done )) Savefolder=evaluated_results '/ ' $modelfolderĭatasets=( 'bottle ' 'cable ' 'capsule ' 'carpet ' 'grid ' 'hazelnut ' 'leather ' 'metal_nut ' 'pill ' 'screw ' 'tile ' 'toothbrush ' 'transistor ' 'wood ' 'zipper ')ĭataset_flags=( $(for dataset in " " do echo '-d ' $dataset done )) Loadpath=/path_to_pretrained_patchcores_models Given a pretrained PatchCore model (or models for all MVTec AD subdatasets), these can be evaluated using For other sample runs with different backbones, larger images or ensembles, see Which runs PatchCore on MVTec images of sizes 224x224 using a WideResNet50-backbone pretrained on Sampler -p 0.1 approx_greedy_coreset dataset -resize 256 -imagesize 224 mvtec $datapath
pretrain_embed_dimension 1024 -target_embed_dimension 1024 -anomaly_scorer_num_nn 1 -patchsize 3 \ Patch_core -b wideresnet50 -le layer2 -le layer3 -faiss_on_gpu \ Python bin/run_patchcore.py -gpu 0 -seed 0 -save_patchcore_model \ 'leather' 'metal_nut' 'pill' 'screw' 'tile' 'toothbrush' 'transistor' 'wood' 'zipper')ĭataset_flags=($(for dataset in do echo '-d '$dataset done)) Datapath=/path_to_mvtec_folder/mvtec datasets=('bottle' 'cable' 'capsule' 'carpet' 'grid' 'hazelnut'
0 kommentar(er)
