GIT: A Generative Image-to-text Transformer for Vision and Language

Benchmark Model Rank Results
image-captioning-on-coco-captionsGIT#3BLEU-4: 44.1CIDER: 151.1METEOR: 32.2SPICE: 26.3
image-captioning-on-nocaps-entireGIT, Single Model#1CIDEr: 123.39B1: 88.1B2: 74.81B3: 57.68B4: 37.35
image-captioning-on-nocaps-in-domainGIT2, Single Model#2CIDEr: 124.18B1: 88.86B2: 75.86B3: 59.94B4: 41.1
image-captioning-on-nocaps-in-domainGIT, Single Model#3CIDEr: 122.4B1: 88.55B2: 76.1B3: 60.53B4: 41.65
image-captioning-on-nocaps-near-domainGIT2, Single Model#1CIDEr: 125.51B1: 88.9B2: 75.86B3: 58.9B4: 38.95
image-captioning-on-nocaps-near-domainGIT, Single Model#3CIDEr: 123.92B1: 88.56B2: 75.48B3: 58.46B4: 38.44
image-captioning-on-nocaps-out-of-domainGIT2, Single Model#2CIDEr: 122.27B1: 86.28B2: 71.15B3: 52.36B4: 30.15
image-captioning-on-nocaps-out-of-domainGIT, Single Model#3CIDEr: 122.04B1: 85.99B2: 71.28B3: 52.66B4: 30.04
image-captioning-on-nocaps-xd-entireGIT2#1CIDEr: 124.77B1: 88.43B2: 75.02B3: 57.87B4: 37.65
image-captioning-on-nocaps-xd-entireGIT#2CIDEr: 123.39B1: 88.1B2: 74.81B3: 57.68B4: 37.35
image-captioning-on-nocaps-xd-in-domainGIT2#1CIDEr: 124.18B1: 88.86B2: 75.86B3: 59.94B4: 41.1
image-captioning-on-nocaps-xd-in-domainGIT#2CIDEr: 122.4B1: 88.55B2: 76.1B3: 60.53B4: 41.65
image-captioning-on-nocaps-xd-near-domainGIT2#1CIDEr: 125.51B1: 88.9B2: 75.86B3: 58.9B4: 38.95
image-captioning-on-nocaps-xd-near-domainGIT#2CIDEr: 123.92B1: 88.56B2: 75.48B3: 58.46B4: 38.44
image-captioning-on-nocaps-xd-out-of-domainGIT2#1CIDEr: 122.27B1: 86.28B2: 71.15B3: 52.36B4: 30.15
image-captioning-on-nocaps-xd-out-of-domainGIT#2CIDEr: 122.04B1: 85.99B2: 71.28B3: 52.66B4: 30.04
video-captioning-on-msr-vtt-1GIT2#3CIDEr: 75.9METEOR: 33.1ROUGE-L: 68.2BLEU-4: 54.8GS: 201.6
visual-question-answering-on-msvd-qa-1GIT#7Accuracy: 0.568