| A | B | C | D | E | F | G | H | I | J | K | L | M | N | O | P | Q | R | S | T | U | V | W | X | Y | Z | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1 | pretrain | link | Pretrain data | Note | Approve | |||||||||||||||||||||
2 | ViT (Vision Transformer) bản base | https://huggingface.co/google/vit-base-patch16-224 | Image net | Accepted | ||||||||||||||||||||||
3 | ViT/base pretrained trên tập Imagenet theo paper DeIT: Training data-efficient image transformers & distillation through attention | https://dl.fbaipublicfiles.com/deit/deit_base_patch16_224-b5f2ef4d.pth | Image net | Accepted | ||||||||||||||||||||||
4 | Text image super resolution : https://github.com/mjq11302010044/Real-CE/tree/main | https://drive.google.com/file/d/1wga0xFdBSkAt_Pif3wPMG4tnHA9wQ7wD/view?usp=sharing | Image net | super resolution problem | Accepted | |||||||||||||||||||||
5 | backbone VGG19 được train trên tập IMAGENET 1K với nhiệm vụ phân loại | https://download.pytorch.org/models/vgg19_bn-c79401a0.pth | Image net | Accepted | ||||||||||||||||||||||
6 | ABINet: Read Like Humans: Autonomous, Bidirectional and Iterative Language Modeling for Scene Text Recognition. Using MJSynth and SynthText two text recognition datasets for training, and evaluating on IIIT, SVT, IC03, IC13, IC15, SVTP, CUTE datasets | https://paddleocr.bj.bcebos.com/rec_r45_abinet_train.tar | pretrain trên synthtext và Mjsynth | Dữ liệu synthetic | Accepted | |||||||||||||||||||||
7 | VGG19 | https://pytorch.org/vision/main/models/generated/torchvision.models.vgg19_bn.html | Image net | Accepted | ||||||||||||||||||||||
8 | Vgg19 | https://pytorch.org/vision/stable/models/generated/torchvision.models.vgg19_bn.html#torchvision.models.VGG19_BN_Weights | Image net | Accepted | ||||||||||||||||||||||
9 | ViT/Base được train trên tập Imagenet theo paper DeIT: Training data-efficient image transformers & distillation through attention | https://dl.fbaipublicfiles.com/deit/deit_base_patch16_224-b5f2ef4d.pth | Image net | Accepted | ||||||||||||||||||||||
10 | Model nhận diện chữ của Clova AI Research nổi tiếng vào 2019 | https://drive.google.com/file/d/1b59rXuGGmKne1AuHnkgDzoYgKeETNMv9/view?usp=sharing | Pretrain MJSynth (MJ)[1], SynthText (ST) | Dữ liệu synthetic | Accepted | |||||||||||||||||||||
11 | Pretrain CLIP dùng làm backbone | https://huggingface.co/timm/convnext_large_mlp.clip_laion2b_soup_ft_in12k_in1k_384 | Image net | Accepted | ||||||||||||||||||||||
12 | vgg pretrained với Imagenet Dataset | https://pytorch.org/vision/stable/models/vgg.html | Image net | Accepted | ||||||||||||||||||||||
13 | ABINet | https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.6/doc/doc_en/algorithm_rec_abinet_en.md | pretrain trên synthtext và Mjsynth | Dữ liệu synthetic | Accepted | |||||||||||||||||||||
14 | Mô hình đa ngôn ngữ XLM Roberta bản base | https://huggingface.co/xlm-roberta-base | Image net | Language model | Accepted | |||||||||||||||||||||
15 | ViT/small pretrained trên tập Imagenet theo paper DeIT: Training data-efficient image transformers & distillation through attention | https://dl.fbaipublicfiles.com/deit/deit_small_patch16_224-cd65a155.pth | Image net | Accepted | ||||||||||||||||||||||
16 | Text image super resolution :https://github.com/csxmli2016/textbsr | https://github.com/csxmli2016/textbsr/releases/download/0.2.0/bsrgan_text_256.pth | Image net | super resolution problem | Accepted | |||||||||||||||||||||
17 | Reciprocal Feature Learning via Explicit and Implicit Tasks in Scene Text Recognition. Using MJSynth and SynthText two text recognition datasets for training, and evaluating on IIIT, SVT, IC03, IC13, IC15, SVTP, CUTE datasets. | https://paddleocr.bj.bcebos.com/contribution/rec_resnet_rfl_att_train.tar | pretrain trên synthtext và Mjsynth | Dữ liệu synthetic | Accepted | |||||||||||||||||||||
18 | ResNet | https://pytorch.org/vision/stable/models/resnet.html | Image net | Accepted | ||||||||||||||||||||||
19 | RESNET50 | https://pytorch.org/vision/master/models/generated/torchvision.models.resnet50.html | Image net | Accepted | ||||||||||||||||||||||
20 | PaddleOCR | https://github.com/PaddlePaddle/PaddleOCR | pretrain trên synthtext và Mjsynth | Dữ liệu synthetic | Accepted | |||||||||||||||||||||
21 | ViT/Small được train trên tập Imagenet theo paper DeIT: Training data-efficient image transformers & distillation through attention | https://dl.fbaipublicfiles.com/deit/deit_small_patch16_224-cd65a155.pth | Imagenet | Accepted | ||||||||||||||||||||||
22 | pretrain clip dùng làm backbone | https://huggingface.co/timm/convnext_large_mlp.clip_laion2b_soup_ft_in12k_in1k_320 | Image net | Accepted | ||||||||||||||||||||||
23 | Resnet pretrained với Imagenet Dataset | Image net | Accepted | |||||||||||||||||||||||
24 | SVTR | https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.6/doc/doc_en/algorithm_rec_svtr_en.md | pretrain trên synthtext và Mjsynth | Dữ liệu synthetic | Accepted | |||||||||||||||||||||
25 | Mô hình pretrained tiếng anh MATRN | https://www.dropbox.com/s/pjcarm73cqwbxh4/best-train-matrn.pth?dl=0 | Bộ dữ liệu synthetic: Synthtext, Mjsynth, wikitext | Dữ liệu synthetic | Accepted | |||||||||||||||||||||
26 | From Two to One: A New Scene Text Recognizer with Visual Language Modeling Network. Using MJSynth and SynthText two text recognition datasets for training, and evaluating on IIIT, SVT, IC13, IC15, SVTP, CUTE datasets. | https://paddleocr.bj.bcebos.com/VisionLAN/rec_r45_visionlan_train.tar | pretrain trên synthtext và Mjsynth | Dữ liệu synthetic | Accepted | |||||||||||||||||||||
27 | ViTSTR là mô hình một giai đoạn đơn giản sử dụng Vision Transformer (ViT) được đào tạo trước để thực hiện Nhận dạng văn bản cảnh (ViTSTR). Nó có độ chính xác tương đương với các mô hình STR hiện đại mặc dù nó sử dụng số lượng tham số và FLOPS ít hơn đáng kể. ViTSTR cũng nhanh do tính toán song song vốn có của kiến trúc ViT. | https://github.com/roatienza/deep-text-recognition-benchmark | synthetic training datasets MJSynth (MJ) and SynthText | Dữ liệu synthetic | Accepted | |||||||||||||||||||||
28 | YOLO | https://github.com/ultralytics/ultralytics | Image net | Accepted | ||||||||||||||||||||||
29 | ViT/Base train trên tập dữ liệu tiếng anh theo bài ViTSTR: Vision Transformer for Fast and Efficient Scene Text Recognition | https://github.com/roatienza/deep-text-recognition-benchmark/releases/download/v0.1.0/vitstr_base_patch16_224_aug.pth | synthetic training datasets MJSynth (MJ) and SynthText | Dữ liệu synthetic | Accepted | |||||||||||||||||||||
30 | Swin Transformer (large-sized model) | https://huggingface.co/microsoft/swin-base-patch4-window12-384-in22k | Image net | Accepted | ||||||||||||||||||||||
31 | ViT/Small train trên tập dữ liệu tiếng anh theo bài ViTSTR: Vision Transformer for Fast and Efficient Scene Text Recognition | https://github.com/roatienza/deep-text-recognition-benchmark/releases/download/v0.1.0/vitstr_small_patch16_224_aug.pth | synthetic training datasets MJSynth (MJ) and SynthText | Dữ liệu synthetic | Accepted | |||||||||||||||||||||
32 | SVTR: Scene Text Recognition with a Single Visual Model (SVTR Large) | https://paddleocr.bj.bcebos.com/PP-OCRv3/chinese/rec_svtr_large_none_ctc_en_train.tar | Pretrain on image net | Accepted | ||||||||||||||||||||||
33 | BEiT (base-sized model, fine-tuned on ImageNet-22k) | https://huggingface.co/microsoft/beit-base-patch16-224-pt22k-ft22k | Image net | Accepted | ||||||||||||||||||||||
34 | pretrain backbone | https://huggingface.co/timm/mobilenetv3_large_100.ra_in1k | Image net | Accepted | ||||||||||||||||||||||
35 | YOLOS (base-sized) model | https://huggingface.co/hustvl/yolos-base | Pretrain Imagenet Finetune: COCO detection | Accepted | ||||||||||||||||||||||
36 | ABI net | https://paddleocr.bj.bcebos.com/rec_r45_abinet_train.tar | synthetic training datasets MJSynth (MJ) and SynthText | Dữ liệu synthetic | Accepted | |||||||||||||||||||||
37 | SVTR | https://paddleocr.bj.bcebos.com/PP-OCRv3/chinese/rec_svtr_large_none_ctc_en_train.tar | synthetic training datasets MJSynth (MJ) and SynthText | Dữ liệu synthetic | Accepted | |||||||||||||||||||||
38 | pretrain backbone tổng hợp tại thư viện timm | https://huggingface.co/timm | Image net | Accepted | ||||||||||||||||||||||
39 | SRN | https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.6/doc/doc_en/algorithm_rec_srn_en.md | pretrain trên synthtext và Mjsynth | Dữ liệu synthetic | Accepted | |||||||||||||||||||||
40 | VGG Model | https://pytorch.org/vision/main/models/vgg.html | Image net | Accepted | ||||||||||||||||||||||
41 | Vision Transformer (base-sized model) | https://huggingface.co/google/vit-base-patch16-224 | Image net | Accepted | ||||||||||||||||||||||
42 | pretrain VIT dùng làm backbone | timm/vit_large_patch14_clip_224.openai_ft_in12k_in1k | Image net | Accepted | ||||||||||||||||||||||
43 | Efficientnet pretrained với Imagenet Dataset | https://pytorch.org/vision/stable/models/efficientnet.html | Image net | Accepted | ||||||||||||||||||||||
44 | ResNet | https://paperswithcode.com/method/resnet | Imagenet | Accepted | ||||||||||||||||||||||
45 | pretrain backbone | https://huggingface.co/timm/resnet50.a1_in1k | Image net | Accepted | ||||||||||||||||||||||
46 | Vit-Transformer pretrain với cifar, imagenet dataset | https://github.com/google-research/vision_transformer | Cifar, imagenet | Accepted | ||||||||||||||||||||||
47 | VGG19 bn | https://pytorch.org/vision/main/models/generated/torchvision.models.vgg19_bn.html | Image net | Accepted | ||||||||||||||||||||||
48 | Mô hình pretrained tiếng anh ABINet++ | https://drive.google.com/file/d/1p6Pw053fFtwmOWd7Qiw3w4qYKf13-bDg/view?usp=share_link | Dữ liệu pretrain này huấn luyện trên synthtext và Mjtext | Sử dụng synthetic text | Accepted | |||||||||||||||||||||
49 | Các mô hình pretrained tiếng anh của MMOCR như: abinet, satrn, sar, master.... | https://github.com/open-mmlab/mmocr | Các model pretrain được huấn luyện trên các bộ dataset liên quan đến tác vụ OCR tuỳ weight sẽ chỉ dùng synthetic data hoặc data thu thập thuần) | Được sử dụng pretrain của - ABInet - SATRN - SVTR - NRTR - MASTER - ASTER - CRNN | Accepted | |||||||||||||||||||||
50 | SVTR: Scene Text Recognition with a Single Visual Model (SVTR Tiny). | https://paddleocr.bj.bcebos.com/PP-OCRv3/chinese/rec_svtr_tiny_none_ctc_en_train.tar | Train trên dữ liệu synthetic | Accepted | ||||||||||||||||||||||
51 | Mô hình STR parseq | https://github.com/baudm/parseq | synthetic training datasets MJSynth (MJ) [30] and SynthText, Kết hợp với nhiều bộ data text recognition khác COCO Text, UberTextm RCTW17,ART,MLT19,… | Sử dụng dữ liệu ảnh text không phải synthetic trong quá trình huấn luyện: COCO text, Uber Text,... | Rejected | |||||||||||||||||||||
52 | Parseq | https://github.com/baudm/parseq | synthetic training datasets MJSynth (MJ) [30] and SynthText, Kết hợp với nhiều bộ data text recognition khác COCO Text, UberTextm RCTW17,ART,MLT19,… | Sử dụng dữ liệu ảnh text không phải synthetic trong quá trình huấn luyện: COCO text, Uber Text,... | Rejected | |||||||||||||||||||||
53 | nó là pretrained cho model parseq được huấn luyện trên tập dữ liệu tiếng anh cho bài toán scene text do tác giả parseq thực nghiệm để đánh giá các mô hình trong paper parseq | https://github.com/baudm/parseq/releases/download/v1.0.0/parseq-bb5792a6.pt | synthetic training datasets MJSynth (MJ) [30] and SynthText, Kết hợp với nhiều bộ data text recognition khác COCO Text, UberTextm RCTW17,ART,MLT19,… | Sử dụng dữ liệu ảnh text không phải synthetic trong quá trình huấn luyện: COCO text, Uber Text,... | Rejected | |||||||||||||||||||||
54 | Pretrained của mô hình Parseq | https://github.com/baudm/parseq/releases/download/v1.0.0/parseq-bb5792a6.pt | synthetic training datasets MJSynth (MJ) [30] and SynthText, Kết hợp với nhiều bộ data text recognition khác COCO Text, UberTextm RCTW17,ART,MLT19,… | Sử dụng dữ liệu ảnh text không phải synthetic trong quá trình huấn luyện: COCO text, Uber Text,... | Rejected | |||||||||||||||||||||
55 | là với một tập. Mô hình là với một tập hợp các mô hình tự hồi quy (AR), họ có thể thống nhất các phương pháp giải mã STR hiện tại (AR nhận biết theo ngữ cảnh và không phải AR không theo ngữ cảnh) và mô hình sàng lọc hai chiều (cloze). Với tham số hóa bộ giải mã chính xác, nó có thể được huấn luyện bằng Mô hình ngôn ngữ hoán vị để cho phép suy luận về các vị trí đầu ra tùy ý cho các tập hợp con tùy ý của ngữ cảnh đầu vào. Đặc điểm của phương pháp này tạo ra một mô hình STR thống nhất—PARSeq—có khả năng suy luận không ngữ cảnh và nhận biết ngữ cảnh, cũng như sàng lọc dự đoán lặp lại bằng cách sử dụng ngữ cảnh hai chiều mà không yêu cầu mô hình ngôn ngữ độc lập.p hợp các mô hình tự hồi quy (AR), chúng tôi có thể thống nhất các phương pháp giải mã STR hiện tại (AR nhận biết theo ngữ cảnh và không phải AR không theo ngữ cảnh) và mô hình sàng lọc hai chiều (cloze): | https://github.com/baudm/parseq | synthetic training datasets MJSynth (MJ) [30] and SynthText, Kết hợp với nhiều bộ data text recognition khác COCO Text, UberTextm RCTW17,ART,MLT19,… | Sử dụng dữ liệu ảnh text không phải synthetic trong quá trình huấn luyện: COCO text, Uber Text,... | Rejected | |||||||||||||||||||||
56 | The TrOCR model is an encoder-decoder model, consisting of an image Transformer as encoder, and a text Transformer as decoder. The image encoder was initialized from the weights of BEiT, while the text decoder was initialized from the weights of RoBERTa. | https://huggingface.co/microsoft/trocr-base-printed | Finetune on IAM,SROIE Pretrain: Dữ liệu dạng văn bản lấy từ trên mạng( có chứa bộ dataset IIIT-HWS là chữ viết tay) với khá nhiều synthetic data | Có sử dụng IIIT-HWS pretrain là bộ dữ liệu chữ viết tay | Rejected | |||||||||||||||||||||
57 | Parseq-tiny | https://github.com/baudm/parseq/releases/tag/v1.0.0 | synthetic training datasets MJSynth (MJ) [30] and SynthText, Kết hợp với nhiều bộ data text recognition khác COCO Text, UberTextm RCTW17,ART,MLT19,… | Sử dụng dữ liệu ảnh text không phải synthetic trong quá trình huấn luyện: COCO text, Uber Text,... | Rejected | |||||||||||||||||||||
58 | Mô hình CRNN kết hợp vgg19 và transformer | https://vocr.vn/data/vietocr/config/vgg-transformer.yml | Train trên 10M ảnh chữ trên tác vụ text recognition | Pretrain trên bộ data chữ tiếng Việt | Rejected | |||||||||||||||||||||
59 | VietOCR - mô hình kết hợp giữa mô hình CNN và Transformer | https://github.com/pbcquoc/vietocr | Train trên 10M ảnh chữ trên tác vụ text recognition | Pretrain trên bộ data chữ tiếng Việt | Rejected | |||||||||||||||||||||
60 | nó là pretrained cho model abinet được huấn luyện trên tập dữ liệu tiếng anh cho bài toán scene text do tác giả parseq thực nghiệm để đánh giá các mô hình trong paper parseq | https://github.com/baudm/parseq/releases/download/v1.0.0/abinet-1d1e373e.pt | synthetic training datasets MJSynth (MJ) [30] and SynthText, Kết hợp với nhiều bộ data text recognition khác COCO Text, UberTextm RCTW17,ART,MLT19,… | Sử dụng dữ liệu ảnh text không phải synthetic trong quá trình huấn luyện: COCO text, Uber Text,... | Rejected | |||||||||||||||||||||
61 | Mô hình pre-trained tiếng anh của Parseq | https://github.com/baudm/parseq/releases/download/v1.0.0/parseq-bb5792a6.pt | synthetic training datasets MJSynth (MJ) [30] and SynthText, Kết hợp với nhiều bộ data text recognition khác COCO Text, UberTextm RCTW17,ART,MLT19,… | Sử dụng dữ liệu ảnh text không phải synthetic trong quá trình huấn luyện: COCO text, Uber Text,... | Rejected | |||||||||||||||||||||
62 | TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models | https://github.com/microsoft/unilm/tree/master/trocr | Kết hợp synthetic vs collected data | Các bản pretrain đã được train trên dữ liệu text không phải synthetic | Rejected | |||||||||||||||||||||
63 | Pretrained của mô hình ABInet | https://github.com/baudm/parseq/releases/download/v1.0.0/abinet-1d1e373e.pt | synthetic training datasets MJSynth (MJ) [30] and SynthText, Kết hợp với nhiều bộ data text recognition khác COCO Text, UberTextm RCTW17,ART,MLT19,… | Sử dụng dữ liệu ảnh text không phải synthetic trong quá trình huấn luyện: COCO text, Uber Text,... | Rejected | |||||||||||||||||||||
64 | vietocr là mô hình cài đặt mô hình Transformer OCR nhận dạng chữ viết tay, chữ đánh máy cho Tiếng Việt. Kiến trúc mô hình là sự kết hợp tuyệt vời giữ mô hình CNN và Transformer (là mô hình nền tảng của BERT khá nổi tiếng). | https://github.com/pbcquoc/vietocr | Train trên 10M ảnh chữ trên tác vụ text recognition | Pretrain trên bộ data chữ tiếng Việt | Rejected | |||||||||||||||||||||
65 | Model OCR transformer finetune trên chữ viết tay tiếng anh của Microsoft | https://huggingface.co/microsoft/trocr-base-handwritten | Đã được finetune trên bộ chữ tiếng Anh | Các bản pretrain đã được train trên dữ liệu chữ viết tay | Rejected | |||||||||||||||||||||
66 | Parseq | https://github.com/baudm/parseq/releases/tag/v1.0.0 | synthetic training datasets MJSynth (MJ) [30] and SynthText, Kết hợp với nhiều bộ data text recognition khác COCO Text, UberTextm RCTW17,ART,MLT19,… | Sử dụng dữ liệu ảnh text không phải synthetic trong quá trình huấn luyện: COCO text, Uber Text,... | Rejected | |||||||||||||||||||||
67 | nó là pretrained cho model vitstr được huấn luyện trên tập dữ liệu tiếng anh cho bài toán scene text do tác giả parseq thực nghiệm để đánh giá các mô hình trong paper parseq | https://github.com/baudm/parseq/releases/download/v1.0.0/vitstr-26d0fcf4.pt | synthetic training datasets MJSynth (MJ) [30] and SynthText, Kết hợp với nhiều bộ data text recognition khác COCO Text, UberTextm RCTW17,ART,MLT19,… | Sử dụng dữ liệu ảnh text không phải synthetic trong quá trình huấn luyện: COCO text, Uber Text,... | Rejected | |||||||||||||||||||||
68 | Pretrained của mô hình TRBA | https://github.com/baudm/parseq/releases/download/v1.0.0/trba-cfaed284.pt | synthetic training datasets MJSynth (MJ) [30] and SynthText, Kết hợp với nhiều bộ data text recognition khác COCO Text, UberTextm RCTW17,ART,MLT19,… | Sử dụng dữ liệu ảnh text không phải synthetic trong quá trình huấn luyện: COCO text, Uber Text,... | Rejected | |||||||||||||||||||||
69 | parseq_small_patch16_224 | https://github.com/baudm/parseq/releases/tag/v1.0.0 | synthetic training datasets MJSynth (MJ) [30] and SynthText, Kết hợp với nhiều bộ data text recognition khác COCO Text, UberTextm RCTW17,ART,MLT19,… | Sử dụng dữ liệu ảnh text không phải synthetic trong quá trình huấn luyện: COCO text, Uber Text,... | Rejected | |||||||||||||||||||||
70 | Pretrained của mô hình ViSTR | https://github.com/baudm/parseq/releases/download/v1.0.0/vitstr-26d0fcf4.pt | synthetic training datasets MJSynth (MJ) [30] and SynthText, Kết hợp với nhiều bộ data text recognition khác COCO Text, UberTextm RCTW17,ART,MLT19,… | Sử dụng dữ liệu ảnh text không phải synthetic trong quá trình huấn luyện: COCO text, Uber Text,... | Rejected | |||||||||||||||||||||
71 | Pretrained của mô hình CRNN | https://github.com/baudm/parseq/releases/download/v1.0.0/crnn-679d0e31.pt | synthetic training datasets MJSynth (MJ) [30] and SynthText, Kết hợp với nhiều bộ data text recognition khác COCO Text, UberTextm RCTW17,ART,MLT19,… | Sử dụng dữ liệu ảnh text không phải synthetic trong quá trình huấn luyện: COCO text, Uber Text,... | Rejected | |||||||||||||||||||||
72 | Backbone gồm resnet, transformer kết hợp với position attention | https://awscv-public-data.s3.us-west-2.amazonaws.com/semimtr/semimtr_vision_model_real_l_and_u.pth | Không tra được dữ liệu pretrain | Rejected | ||||||||||||||||||||||
73 | Một biến thể của khối decoder trong mô hình transformer | https://awscv-public-data.s3.us-west-2.amazonaws.com/semimtr/abinet_language_model.pth | Không tra được dữ liệu pretrain | Rejected | ||||||||||||||||||||||
74 | ||||||||||||||||||||||||||
75 | ||||||||||||||||||||||||||
76 | ||||||||||||||||||||||||||
77 | ||||||||||||||||||||||||||
78 | ||||||||||||||||||||||||||
79 | ||||||||||||||||||||||||||
80 | ||||||||||||||||||||||||||
81 | ||||||||||||||||||||||||||
82 | ||||||||||||||||||||||||||
83 | ||||||||||||||||||||||||||
84 | ||||||||||||||||||||||||||
85 | ||||||||||||||||||||||||||
86 | ||||||||||||||||||||||||||
87 | ||||||||||||||||||||||||||
88 | ||||||||||||||||||||||||||
89 | ||||||||||||||||||||||||||
90 | ||||||||||||||||||||||||||
91 | ||||||||||||||||||||||||||
92 | ||||||||||||||||||||||||||
93 | ||||||||||||||||||||||||||
94 | ||||||||||||||||||||||||||
95 | ||||||||||||||||||||||||||
96 | ||||||||||||||||||||||||||
97 | ||||||||||||||||||||||||||
98 | ||||||||||||||||||||||||||
99 | ||||||||||||||||||||||||||
100 |