iflytek / VLE

VLE: Vision-Language Encoder (VLE: 视觉-语言多模态预训练模型)
Apache License 2.0
176 stars 12 forks source link

not all images have caption annotations #4

Closed zhousteven closed 1 year ago

zhousteven commented 1 year ago

when i run script "write_vqa.py", use the public datasets as author recommended in "README.md", console output some message as show below: ./VLE/VLE-main/examples/VQA/write_vqa.py 100%|██████████| 443757/443757 [00:00<00:00, 851830.30it/s] 100%|██████████| 214354/214354 [00:00<00:00, 431325.80it/s] 100%|██████████| 447793/447793 [00:00<00:00, 809395.66it/s] 100%|██████████| 107394/107394 [00:00<00:00, 3530035.22it/s] 100%|██████████| 443757/443757 [00:00<00:00, 7974002.36it/s] 100%|██████████| 214354/214354 [00:00<00:00, 7733833.17it/s] 100%|██████████| 658111/658111 [00:06<00:00, 100919.83it/s] 100%|██████████| 443757/443757 [00:01<00:00, 359162.75it/s] 100%|██████████| 214354/214354 [00:00<00:00, 237090.17it/s] not all images have caption annotations 82783 82774 82774 100%|██████████| 82774/82774 [56:48<00:00, 24.28it/s] not all images have caption annotations 40504 40503 40503 100%|██████████| 40503/40503 [26:37<00:00, 25.35it/s] 0%| | 0/81434 [00:00<?, ?it/s]all images have caption annotations 81434 81434 81434 100%|██████████| 81434/81434 [49:52<00:00, 27.21it/s] not all images have caption annotations 81434 36807 36807 100%|██████████| 36807/36807 [19:12<00:00, 31.93it/s]

Process finished with exit code 0

some line mentioned that" not all image have caption annotations",should i care about this message? whether it effect the next training steps? please kindly answer my doubts~ tks~

GoGoJoestar commented 1 year ago

The message only means some images have no annotations in this dataset. It is normal in Dataset Preparation process.