Hi, I am trying to train LSVT dataset with this repo.

Seems, you've been trained LSVT, https://github.com/mxin262/SwinTextSpotter/blob/e238a4b5d0c127480a838c6245c1e5e9eb2f9d59/detectron2/data/datasets/builtin.py#L58 but there is no transcription for that.

is LSVT data trainable with this code? then can you provide config files and lexicon for that?

if I use the ctw1500 dataset, how to get the annotation file " instances_train2017.json" and "test_ctw1500_maxlen100.json", thanks.

Parameter setting： weights: rects_model_final.pth config: ./projects/SWINTS/configs/SWINTS-swin-chn_finetune.yaml predect data: ReCTS/icdar2019_rects_images1/* run script: python3 demo/demo.py --opts MODEL.WEIGHTS ./weights/rects_model_final.pth

Using the above parameters does not predict results, but 'tt_model_final.pth' is right

Thank you for your kindly code sharing!

And when I run : python projects/SWINTS/train_net.py --config-file projects/SWINTS/configs/SWINTS-swin-finetune-totaltext.yaml --eval-only MODEL.WEIGHTS model.pth

The error occurs: FileNotFoundError: [Errno 2] No such file or directory: 'datasets/totaltext/weak_voc_new.txt'

But I can not find this file according to dataset download.

I want to train SwinTextSpotter on my custom dataset (in totaltext format). But when I start to train, I was facing with this error: "AttributeError: Cannot find field 'gt_masks' in the given Instances!" in line 230 of file projects/SWINTS/swints/swints.py How can I get gt_masks for my custom dataset?

Thank you for your hard work!

执行如下指令：python projects/SWINTS/train_net.py --num-gpus 4 --config-file projects/SWINTS/configs/SWINTS-swin-pretrain.yaml

4 A100 80G batchsize=24，rec loss始终在6-7左右

[08/03 14:38:52] detectron2 INFO: Rank of current process: 0. World size: 4 [08/03 14:38:53] detectron2 INFO: Environment info:

sys.platform linux Python 3.8.1 (default, Jan 8 2020, 22:29:32) [GCC 7.3.0] numpy 1.23.1 detectron2 0.4 @/home/SwinTextSpotter/detectron2 Compiler GCC 7.5 CUDA compiler CUDA 11.3 detectron2 arch flags 6.1 DETECTRON2_ENV_MODULE PyTorch 1.10.0+cu113 @/miniconda/lib/python3.8/site-packages/torch PyTorch debug build False GPU available True GPU 0,1,2,3 NVIDIA A100 80GB PCIe (arch=8.0) CUDA_HOME /usr/local/cuda Pillow 9.2.0 torchvision 0.11.0+cu113 @/miniconda/lib/python3.8/site-packages/torchvision torchvision arch flags 3.5, 5.0, 6.0, 7.0, 7.5, 8.0, 8.6 fvcore 0.1.5.post20220512 iopath 0.1.7 cv2 4.6.0

PyTorch built with:

• GCC 7.3
• C++ Version: 201402
• Intel(R) Math Kernel Library Version 2020.0.0 Product Build 20191122 for Intel(R) 64 architecture applications
• Intel(R) MKL-DNN v2.2.3 (Git Hash 7336ca9f055cf1bfa13efb658fe15dc9b41f0740)
• OpenMP 201511 (a.k.a. OpenMP 4.5)
• LAPACK is enabled (usually provided by MKL)
• NNPACK is enabled
• CPU capability usage: AVX512
• CUDA Runtime 11.3
• NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86
• CuDNN 8.2
• Magma 2.5.2
• Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=11.3, CUDNN_VERSION=8.2.0, CXX_COMPILER=/opt/rh/devtoolset-7/root/usr/bin/c++, CXX_FLAGS= -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -fopenmp -DNDEBUG -DUSE_KINETO -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -DEDGE_PROFILER_USE_KINETO -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-sign-compare -Wno-unused-parameter -Wno-unused-variable -Wno-unused-function -Wno-unused-result -Wno-unused-local-typedefs -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-psabi -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=1.10.0, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON,

[08/03 14:38:53] detectron2 INFO: Command line arguments: Namespace(config_file='projects/SWINTS/configs/SWINTS-swin-pretrain.yaml', dist_url='tcp://127.0.0.1:49152', eval_only=False, machine_rank=0, num_gpus=4, num_machines=1, opts=[], resume=False) [08/03 14:38:53] detectron2 INFO: Contents of args.config_file=projects/SWINTS/configs/SWINTS-swin-pretrain.yaml: BASE: "Base-SWINTS_swin.yaml" MODEL: WEIGHTS: "ckpt/swin_imagenet_pretrain.pth" SWINTS: NUM_PROPOSALS: 300 NUM_CLASSES: 2 DATASETS: TRAIN: ("totaltext_train","icdar_2015_train","icdar_2013_train","icdar_2017_validation_mlt","icdar_2017_mlt","icdar_curvesynthtext_train1","icdar_curvesynthtext_train2",) TEST: ("totaltext_test",) SOLVER: STEPS: (120000,140000) MAX_ITER: 150000 CHECKPOINT_PERIOD: 5000 INPUT: FORMAT: "RGB"

[08/03 14:38:53] detectron2 INFO: Running with full config: CUDNN_BENCHMARK: False DATALOADER: ASPECT_RATIO_GROUPING: True FILTER_EMPTY_ANNOTATIONS: True NUM_WORKERS: 4 REPEAT_THRESHOLD: 0.0 SAMPLER_TRAIN: TrainingSampler DATASETS: PRECOMPUTED_PROPOSAL_TOPK_TEST: 1000 PRECOMPUTED_PROPOSAL_TOPK_TRAIN: 2000 PROPOSAL_FILES_TEST: () PROPOSAL_FILES_TRAIN: () TEST: ('totaltext_test',) TRAIN: ('totaltext_train', 'icdar_2015_train', 'icdar_2013_train', 'icdar_2017_validation_mlt', 'icdar_2017_mlt', 'icdar_curvesynthtext_train1', 'icdar_curvesynthtext_train2') GLOBAL: HACK: 1.0 INPUT: CROP: CROP_INSTANCE: False ENABLED: True SIZE: [0.1, 0.1] TYPE: relative_range FORMAT: RGB MASK_FORMAT: polygon MAX_SIZE_TEST: 1824 MAX_SIZE_TRAIN: 1600 MIN_SIZE_TEST: 1000 MIN_SIZE_TRAIN: (640, 672, 704, 736, 768, 800, 832, 864, 896) MIN_SIZE_TRAIN_SAMPLING: choice RANDOM_FLIP: horizontal MODEL: ANCHOR_GENERATOR: ANGLES: [[-90, 0, 90]] ASPECT_RATIOS: [[0.5, 1.0, 2.0]] NAME: DefaultAnchorGenerator OFFSET: 0.0 SIZES: [[32, 64, 128, 256, 512]] BACKBONE: FREEZE_AT: -1 NAME: build_swint_fpn_backbone DEVICE: cuda FPN: FUSE_TYPE: sum IN_FEATURES: ['stage2', 'stage3', 'stage4', 'stage5'] NORM: OUT_CHANNELS: 256 TOP_LEVELS: 2 KEYPOINT_ON: False LOAD_PROPOSALS: False MASK_ON: True META_ARCHITECTURE: SWINTS PANOPTIC_FPN: COMBINE: ENABLED: True INSTANCES_CONFIDENCE_THRESH: 0.5 OVERLAP_THRESH: 0.5 STUFF_AREA_LIMIT: 4096 INSTANCE_LOSS_WEIGHT: 1.0 PIXEL_MEAN: [123.675, 116.28, 103.53] PIXEL_STD: [58.395, 57.12, 57.375] PROPOSAL_GENERATOR: MIN_SIZE: 0 NAME: RPN REC_HEAD: BATCH_SIZE: 128 NUM_CLASSES: 107 POOLER_RESOLUTION: (28, 28) RESOLUTION: (32, 32) RESNETS: DEFORM_MODULATED: False DEFORM_NUM_GROUPS: 1 DEFORM_ON_PER_STAGE: [False, False, False, False] DEPTH: 50 NORM: FrozenBN NUM_GROUPS: 1 OUT_FEATURES: ['res4'] RES2_OUT_CHANNELS: 256 RES5_DILATION: 1 STEM_OUT_CHANNELS: 64 STRIDE_IN_1X1: True WIDTH_PER_GROUP: 64 RETINANET: BBOX_REG_LOSS_TYPE: smooth_l1 BBOX_REG_WEIGHTS: (1.0, 1.0, 1.0, 1.0) FOCAL_LOSS_ALPHA: 0.25 FOCAL_LOSS_GAMMA: 2.0 IN_FEATURES: ['p3', 'p4', 'p5', 'p6', 'p7'] IOU_LABELS: [0, -1, 1] IOU_THRESHOLDS: [0.4, 0.5] NMS_THRESH_TEST: 0.5 NORM: NUM_CLASSES: 80 NUM_CONVS: 4 PRIOR_PROB: 0.01 SCORE_THRESH_TEST: 0.05 SMOOTH_L1_LOSS_BETA: 0.1 TOPK_CANDIDATES_TEST: 1000 ROI_BOX_CASCADE_HEAD: BBOX_REG_WEIGHTS: ((10.0, 10.0, 5.0, 5.0), (20.0, 20.0, 10.0, 10.0), (30.0, 30.0, 15.0, 15.0)) IOUS: (0.5, 0.6, 0.7) ROI_BOX_HEAD: BBOX_REG_LOSS_TYPE: smooth_l1 BBOX_REG_LOSS_WEIGHT: 1.0 BBOX_REG_WEIGHTS: (10.0, 10.0, 5.0, 5.0) CLS_AGNOSTIC_BBOX_REG: False CONV_DIM: 256 FC_DIM: 1024 NAME: NORM: NUM_CONV: 0 NUM_FC: 0 POOLER_RESOLUTION: 7 POOLER_SAMPLING_RATIO: 2 POOLER_TYPE: ROIAlignV2 SMOOTH_L1_BETA: 0.0 TRAIN_ON_PRED_BOXES: False ROI_HEADS: BATCH_SIZE_PER_IMAGE: 512 IN_FEATURES: ['p2', 'p3', 'p4', 'p5'] IOU_LABELS: [0, 1] IOU_THRESHOLDS: [0.5] NAME: Res5ROIHeads NMS_THRESH_TEST: 0.5 NUM_CLASSES: 80 POSITIVE_FRACTION: 0.25 PROPOSAL_APPEND_GT: True SCORE_THRESH_TEST: 0.05 ROI_KEYPOINT_HEAD: CONV_DIMS: (512, 512, 512, 512, 512, 512, 512, 512) LOSS_WEIGHT: 1.0 MIN_KEYPOINTS_PER_IMAGE: 1 NAME: KRCNNConvDeconvUpsampleHead NORMALIZE_LOSS_BY_VISIBLE_KEYPOINTS: True NUM_KEYPOINTS: 17 POOLER_RESOLUTION: 14 POOLER_SAMPLING_RATIO: 0 POOLER_TYPE: ROIAlignV2 ROI_MASK_HEAD: CLS_AGNOSTIC_MASK: False CONV_DIM: 256 NAME: MaskRCNNConvUpsampleHead NORM: NUM_CONV: 0 POOLER_RESOLUTION: 14 POOLER_SAMPLING_RATIO: 0 POOLER_TYPE: ROIAlignV2 RPN: BATCH_SIZE_PER_IMAGE: 256 BBOX_REG_LOSS_TYPE: smooth_l1 BBOX_REG_LOSS_WEIGHT: 1.0 BBOX_REG_WEIGHTS: (1.0, 1.0, 1.0, 1.0) BOUNDARY_THRESH: -1 CONV_DIMS: [-1] HEAD_NAME: StandardRPNHead IN_FEATURES: ['res4'] IOU_LABELS: [0, -1, 1] IOU_THRESHOLDS: [0.3, 0.7] LOSS_WEIGHT: 1.0 NMS_THRESH: 0.7 POSITIVE_FRACTION: 0.5 POST_NMS_TOPK_TEST: 1000 POST_NMS_TOPK_TRAIN: 2000 PRE_NMS_TOPK_TEST: 6000 PRE_NMS_TOPK_TRAIN: 12000 SMOOTH_L1_BETA: 0.0 SEM_SEG_HEAD: COMMON_STRIDE: 4 CONVS_DIM: 128 IGNORE_VALUE: 255 IN_FEATURES: ['p2', 'p3', 'p4', 'p5'] LOSS_WEIGHT: 1.0 NAME: SemSegFPNHead NORM: GN NUM_CLASSES: 54 SWINT: APE: False DEPTHS: [2, 2, 6, 2] DROP_PATH_RATE: 0.2 EMBED_DIM: 96 MLP_RATIO: 4 NUM_HEADS: [3, 6, 12, 24] OUT_FEATURES: ['stage2', 'stage3', 'stage4', 'stage5'] WINDOW_SIZE: 7 SWINTS: ACTIVATION: relu ALPHA: 0.25 CLASS_WEIGHT: 2.0 DEEP_SUPERVISION: True DIM_DYNAMIC: 64 DIM_FEEDFORWARD: 2048 DROPOUT: 0.0 GAMMA: 2.0 GIOU_WEIGHT: 2.0 HIDDEN_DIM: 256 IOU_LABELS: [0, 1] IOU_THRESHOLDS: [0.5] L1_WEIGHT: 5.0 MASK_DIM: 60 MASK_WEIGHT: 2.0 NHEADS: 8 NO_OBJECT_WEIGHT: 0.1 NUM_CLASSES: 2 NUM_CLS: 3 NUM_DYNAMIC: 2 NUM_HEADS: 6 NUM_MASK: 3 NUM_PROPOSALS: 300 NUM_REG: 3 PATH_COMPONENTS: ./projects/SWINTS/LME/coco_2017_train_class_agnosticTrue_whitenTrue_sigmoidTrue_60_siz28.npz PRIOR_PROB: 0.01 REC_WEIGHT: 1.0 TEST_NUM_PROPOSALS: 100 WEIGHTS: ckpt/swin_imagenet_pretrain.pth OUTPUT_DIR: ./output SEED: 40244023 SOLVER: AMP: ENABLED: False BACKBONE_MULTIPLIER: 1.0 BASE_LR: 7.5e-05 BIAS_LR_FACTOR: 1.0 CHECKPOINT_PERIOD: 5000 CLIP_GRADIENTS: CLIP_TYPE: full_model CLIP_VALUE: 1.0 ENABLED: True NORM_TYPE: 2.0 GAMMA: 0.1 IMS_PER_BATCH: 24 LR_SCHEDULER_NAME: WarmupMultiStepLR MAX_ITER: 150000 MOMENTUM: 0.9 NESTEROV: False OPTIMIZER: ADAMW REFERENCE_WORLD_SIZE: 0 STEPS: (120000, 140000) WARMUP_FACTOR: 0.01 WARMUP_ITERS: 1000 WARMUP_METHOD: linear WEIGHT_DECAY: 0.0001 WEIGHT_DECAY_BIAS: 0.0001 WEIGHT_DECAY_NORM: 0.0 TEST: AUG: ENABLED: False FLIP: True MAX_SIZE: 4000 MIN_SIZES: (400, 500, 600, 700, 800, 900, 1000, 1100, 1200) DETECTIONS_PER_IMAGE: 100 EVAL_PERIOD: 100000 EXPECTED_RESULTS: [] INFERENCE_TH_TEST: 0.4 KEYPOINT_OKS_SIGMAS: [] PRECISE_BN: ENABLED: False NUM_ITER: 200 USE_NMS_IN_TSET: True VERSION: 2 VIS_PERIOD: 0 [08/03 14:38:53] detectron2 INFO: Full config saved to ./output/config.yaml [08/03 14:38:55] d2.engine.defaults INFO: Model:

[08/03 14:38:56] d2.data.datasets.coco INFO: Loaded 1255 images in COCO format from datasets/totaltext/totaltext_train.json [08/03 14:38:56] d2.data.datasets.coco INFO: Loaded 1000 images in COCO format from datasets/icdar2015/icdar_2015_train.json [08/03 14:38:56] d2.data.datasets.coco INFO: Loaded 229 images in COCO format from datasets/icdar2013/annotations/icdar_2013.json [08/03 14:38:56] d2.data.datasets.coco INFO: Loaded 1797 images in COCO format from datasets/mlt2017/annotations/icdar_2017_validation_mlt.json [08/03 14:38:57] d2.data.datasets.coco INFO: Loaded 7160 images in COCO format from datasets/mlt2017/annotations/icdar_2017_mlt.json [08/03 14:39:18] d2.data.datasets.coco INFO: Loading datasets/syntext1/annotations/syntext_word_eng.json takes 21.03 seconds. [08/03 14:39:19] d2.data.datasets.coco INFO: Loaded 94723 images in COCO format from datasets/syntext1/annotations/syntext_word_eng.json [08/03 14:39:35] d2.data.datasets.coco INFO: Loading datasets/syntext2/annotations/ecms_v1_maxlen25.json takes 7.61 seconds. [08/03 14:39:35] d2.data.datasets.coco INFO: Loaded 54327 images in COCO format from datasets/syntext2/annotations/ecms_v1_maxlen25.json [08/03 14:39:36] d2.data.build INFO: Removed 2230 images with no usable annotations. 158261 images left. [08/03 14:39:42] d2.data.build INFO: Distribution of instances among all 1 categories: | category | #instances | |:----------:|:-------------| | text | 1868890 | | | | [08/03 14:39:42] d2.data.build INFO: Using training sampler TrainingSampler [08/03 14:39:42] d2.data.common INFO: Serializing 158261 elements to byte tensors and concatenating them all ... [08/03 14:39:48] d2.data.common INFO: Serialized dataset takes 460.55 MiB [08/03 14:39:50] fvcore.common.checkpoint INFO: [Checkpointer] Loading from ckpt/swin_imagenet_pretrain.pth ...

[08/03 14:39:51] d2.engine.train_loop INFO: Starting training from iteration 0

[08/04 14:53:46 d2.utils.events]: eta: 1 day, 14:59:42 iter: 56279 total_loss: 16.8 loss_ce: 0.1334 loss_giou: 0.29 loss_bbox: 0.1194 loss_feat: 0.5878 loss_dice: 0.119 loss_rec: 6.447 loss_ce_0: 0.6154 loss_giou_0: 0.7695 loss_bbox_0: 0.313 loss_feat_0: 1.204 loss_dice_0: 0.2254 loss_ce_1: 0.3727 loss_giou_1: 0.3973 loss_bbox_1: 0.1591 loss_feat_1: 0.8152 loss_dice_1: 0.1459 loss_ce_2: 0.31 loss_giou_2: 0.3258 loss_bbox_2: 0.1341 loss_feat_2: 0.6907 loss_dice_2: 0.1311 loss_ce_3: 0.2053 loss_giou_3: 0.3006 loss_bbox_3: 0.122 loss_feat_3: 0.6204 loss_dice_3: 0.1216 loss_ce_4: 0.1435 loss_giou_4: 0.2943 loss_bbox_4: 0.1182 loss_feat_4: 0.6 loss_dice_4: 0.1188 time: 1.5465 data_time: 0.0930 lr: 7.5e-05 max_mem: 66803M [08/04 14:54:16 d2.utils.events]: eta: 1 day, 14:59:52 iter: 56299 total_loss: 17.43 loss_ce: 0.1425 loss_giou: 0.2743 loss_bbox: 0.1041 loss_feat: 0.5848 loss_dice: 0.1222 loss_rec: 6.74 loss_ce_0: 0.6106 loss_giou_0: 0.7541 loss_bbox_0: 0.3087 loss_feat_0: 1.213 loss_dice_0: 0.2271 loss_ce_1: 0.378 loss_giou_1: 0.3861 loss_bbox_1: 0.1526 loss_feat_1: 0.8113 loss_dice_1: 0.1538 loss_ce_2: 0.3031 loss_giou_2: 0.3231 loss_bbox_2: 0.1218 loss_feat_2: 0.6802 loss_dice_2: 0.1366 loss_ce_3: 0.2059 loss_giou_3: 0.2907 loss_bbox_3: 0.1099 loss_feat_3: 0.6195 loss_dice_3: 0.127 loss_ce_4: 0.1465 loss_giou_4: 0.2804 loss_bbox_4: 0.1056 loss_feat_4: 0.5916 loss_dice_4: 0.123 time: 1.5464 data_time: 0.0578 lr: 7.5e-05 max_mem: 66803M [08/04 14:54:47 d2.utils.events]: eta: 1 day, 14:59:22 iter: 56319 total_loss: 16.86 loss_ce: 0.1324 loss_giou: 0.2875 loss_bbox: 0.1126 loss_feat: 0.5857 loss_dice: 0.1232 loss_rec: 6.414 loss_ce_0: 0.6167 loss_giou_0: 0.7572 loss_bbox_0: 0.3231 loss_feat_0: 1.195 loss_dice_0: 0.2153 loss_ce_1: 0.3705 loss_giou_1: 0.3863 loss_bbox_1: 0.1542 loss_feat_1: 0.8011 loss_dice_1: 0.1507 loss_ce_2: 0.2973 loss_giou_2: 0.32 loss_bbox_2: 0.126 loss_feat_2: 0.6879 loss_dice_2: 0.1377 loss_ce_3: 0.1973 loss_giou_3: 0.2947 loss_bbox_3: 0.1173 loss_feat_3: 0.6282 loss_dice_3: 0.1278 loss_ce_4: 0.1405 loss_giou_4: 0.2881 loss_bbox_4: 0.113 loss_feat_4: 0.5965 loss_dice_4: 0.1255 time: 1.5464 data_time: 0.0538 lr: 7.5e-05 max_mem: 66803M

Hello, i want to use your pretrained model which was fine tuned on vintext to spot Vietnamese. But seems like it's predicting characters in the English alphabet. I don't see where to change the alphabet in the config files, where should I take a look at to modify this?

Hello everyone, I want to change character list to match my custom dataset characters. How can I do this? I have tried to change CTLBELS, chars whenever character list appear in the code but I end up with this errror "Assertion cur_target >= 0 && cur_target < n_classes failed".

Thank you!

The README doesn't show any detail about char dictionary for various datasets. I print embedding size in rects_model_final.pth and tt_model_final.pth, getting 5000+, 107, respectively.

Hi, thanks for your sharing. But I've met some errors and wish you can give more suggestions.

I finished all the installation following the readme.md, then run demo.py  following getting_started.md
`
Traceback (most recent call last):
  File "demo.py", line 14, in <module>
    from predictor import VisualizationDemo
  File "/XXX/SwinTextSpotter/demo/predictor.py", line 13, in <module>
    from detectron2.utils.visualizer_chn import Visualizer as Visualizer_chn
ModuleNotFoundError: No module named 'detectron2.utils.visualizer_chn'
`

I don't understant why “import detectron2.utils.visualizer" is OK, but " import detectron2.utils.visualizer_chn" as well as " import detectron2.utils.visualizer_vintext" can't be done.

你好，我想问一下，我使用rects数据集来测试demo.py的结果，发现输出的output里没有汉字，我测试totaltext上的英文数据集就有显示英文的

以下是我输入的指令：使用的model为作者您提供的rects的pth文件 python demo/demo.py --config-file projects/SWINTS/configs/SWINTS-swin-chn_finetune.yaml --input input1.jpg --output ./output --confidence-threshold 0.4 --opts MODEL.WEIGHTS ./output/rects_model_final.pth

SWINTS-swin-chn_finetune.yaml的配置： BASE: "Base-SWINTS_swin.yaml" MODEL: #WEIGHTS: "detectron2://ImageNetPretrained/torchvision/R-50.pkl" WEIGHTS: "./output/rects_model_final.pth" SWINTS: NUM_PROPOSALS: 300 NUM_CLASSES: 2 REC_HEAD: POOLER_RESOLUTION: (16,40) RESOLUTION: (32, 80) BATCH_SIZE: 128 NUM_CLASSES: 5463 DATASETS: TRAIN: ("rects",) TEST: ("totaltext_test",) SOLVER: STEPS: (140000,160000) MAX_ITER: 180000 CHECKPOINT_PERIOD: 10000 INPUT: FORMAT: "RGB"

测试的图片：rects里的测试图

产生的检测结果：显示检测到了7个实例

output结果：输出的jpg文件无汉字

训练的时候出现错误“RuntimeError: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 7.92 GiB total capacity; 6.76 GiB already allocated; 66.50 MiB free; 6.87 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF" 将BANTCH_SIZE设置为1依旧显示该错误，请问作者有解决的方案吗？