From 0c7c3052a037f4c21ee034f116f70d9ad3a2f00d Mon Sep 17 00:00:00 2001 From: "A. Unique TensorFlower" Date: Tue, 18 Apr 2023 11:46:58 -0700 Subject: [PATCH] Checkpoints added for ResNet-(50, 101, 152), SpineNet-(49, 96, 143) and DeepLabV3+ PiperOrigin-RevId: 525208722 --- official/vision/MODEL_GARDEN.md | 14 +++++++------- official/vision/README.md | 25 +++++++++++++------------ 2 files changed, 20 insertions(+), 19 deletions(-) diff --git a/official/vision/MODEL_GARDEN.md b/official/vision/MODEL_GARDEN.md index c4487d6fb..927561ddf 100644 --- a/official/vision/MODEL_GARDEN.md +++ b/official/vision/MODEL_GARDEN.md @@ -30,9 +30,9 @@ segmentation. | Model | Resolution | Epochs | Top-1 | Top-5 | Download | | ------------ |:-------------:|--------:|--------:|--------:|---------:| | ResNet-50 | 224x224 | 90 | 76.1 | 92.9 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/image_classification/imagenet_resnet50_tpu.yaml) | -| ResNet-50 | 224x224 | 200 | 77.1 | 93.5 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/image_classification/imagenet_resnet50_tpu.yaml) | -| ResNet-101 | 224x224 | 200 | 78.3 | 94.2 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/image_classification/imagenet_resnet101_tpu.yaml) | -| ResNet-152 | 224x224 | 200 | 78.7 | 94.3 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/image_classification/imagenet_resnet152_tpu.yaml) | +| ResNet-50 | 224x224 | 200 | 77.1 | 93.5 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/image_classification/imagenet_resnet50_tpu.yaml) \| [ckpt](https://storage.googleapis.com/tf_model_garden/vision/resnet/resnet-50-i224.tar.gz) | +| ResNet-101 | 224x224 | 200 | 78.3 | 94.2 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/image_classification/imagenet_resnet101_tpu.yaml) \| [ckpt](https://storage.googleapis.com/tf_model_garden/vision/resnet/resnet-101-i224.tar.gz) | +| ResNet-152 | 224x224 | 200 | 78.7 | 94.3 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/image_classification/imagenet_resnet152_tpu.yaml) \| [ckpt](https://storage.googleapis.com/tf_model_garden/vision/resnet/resnet-152-i224.tar.gz) | #### ResNet-RS models trained with various settings @@ -114,9 +114,9 @@ evaluated on [COCO](https://cocodataset.org/) val2017. | Backbone | Resolution | Epochs | FLOPs (B) | Params (M) | Box AP | Download | | ------------ |:-------------:| -------:|--------------:|-----------:|--------:|---------:| -| SpineNet-49 | 640x640 | 500 | 85.4| 28.5 | 44.2 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/retinanet/coco_spinenet49_tpu.yaml) \| [TB.dev](https://tensorboard.dev/experiment/n2UN83TkTdyKZn3slCWulg/#scalars&_smoothingWeight=0)| -| SpineNet-96 | 1024x1024 | 500 | 265.4 | 43.0 | 48.5 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/retinanet/coco_spinenet96_tpu.yaml) \| [TB.dev](https://tensorboard.dev/experiment/n2UN83TkTdyKZn3slCWulg/#scalars&_smoothingWeight=0)| -| SpineNet-143 | 1280x1280 | 500 | 524.0 | 67.0 | 50.0 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/retinanet/coco_spinenet143_tpu.yaml) \| [TB.dev](https://tensorboard.dev/experiment/n2UN83TkTdyKZn3slCWulg/#scalars&_smoothingWeight=0)| +| SpineNet-49 | 640x640 | 500 | 85.4| 28.5 | 44.2 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/retinanet/coco_spinenet49_tpu.yaml) \| [ckpt](https://storage.googleapis.com/tf_model_garden/vision/spinenet/spinenet-49-i640.tar.gz) \| [TB.dev](https://tensorboard.dev/experiment/n2UN83TkTdyKZn3slCWulg/#scalars&_smoothingWeight=0)| +| SpineNet-96 | 1024x1024 | 500 | 265.4 | 43.0 | 48.5 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/retinanet/coco_spinenet96_tpu.yaml) \| [ckpt](https://storage.googleapis.com/tf_model_garden/vision/spinenet/spinenet-96-i1024.tar.gz) \| [TB.dev](https://tensorboard.dev/experiment/n2UN83TkTdyKZn3slCWulg/#scalars&_smoothingWeight=0)| +| SpineNet-143 | 1280x1280 | 500 | 524.0 | 67.0 | 50.0 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/retinanet/coco_spinenet143_tpu.yaml) \| [ckpt](https://storage.googleapis.com/tf_model_garden/vision/spinenet/spinenet-143-i1280.tar.gz) \| [TB.dev](https://tensorboard.dev/experiment/n2UN83TkTdyKZn3slCWulg/#scalars&_smoothingWeight=0)| #### Mobile-size RetinaNet (Trained from scratch): @@ -157,7 +157,7 @@ evaluated on [COCO](https://cocodataset.org/) val2017. | Model | Backbone | Resolution | Steps | mIoU | Download | | ---------- | :----------------: | :--------: | ----: | ---: | --------:| | DeepLabV3 | Dilated Resnet-101 | 512x512 | 30k | 78.7 | | -| DeepLabV3+ | Dilated Resnet-101 | 512x512 | 30k | 79.2 | | +| DeepLabV3+ | Dilated Resnet-101 | 512x512 | 30k | 79.2 | [ckpt](https://storage.googleapis.com/tf_model_garden/vision/deeplabv3plus/dilated-resnet-101-deeplabv3plus.tar.gz) | ### CITYSCAPES diff --git a/official/vision/README.md b/official/vision/README.md index 54d2875e0..a4cf3bf43 100644 --- a/official/vision/README.md +++ b/official/vision/README.md @@ -54,9 +54,9 @@ segmentation. | Model | Resolution | Epochs | Top-1 | Top-5 | Download | | ------------ |:-------------:|--------:|--------:|--------:|---------:| | ResNet-50 | 224x224 | 90 | 76.1 | 92.9 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/image_classification/imagenet_resnet50_tpu.yaml) | -| ResNet-50 | 224x224 | 200 | 77.1 | 93.5 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/image_classification/imagenet_resnet50_tpu.yaml) | -| ResNet-101 | 224x224 | 200 | 78.3 | 94.2 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/image_classification/imagenet_resnet101_tpu.yaml) | -| ResNet-152 | 224x224 | 200 | 78.7 | 94.3 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/image_classification/imagenet_resnet152_tpu.yaml) | +| ResNet-50 | 224x224 | 200 | 77.1 | 93.5 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/image_classification/imagenet_resnet50_tpu.yaml) \| [ckpt](https://storage.googleapis.com/tf_model_garden/vision/resnet/resnet-50-i224.tar.gz) | +| ResNet-101 | 224x224 | 200 | 78.3 | 94.2 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/image_classification/imagenet_resnet101_tpu.yaml) \| [ckpt](https://storage.googleapis.com/tf_model_garden/vision/resnet/resnet-101-i224.tar.gz) | +| ResNet-152 | 224x224 | 200 | 78.7 | 94.3 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/image_classification/imagenet_resnet152_tpu.yaml) [ckpt](https://storage.googleapis.com/tf_model_garden/vision/resnet/resnet-152-i224.tar.gz) | @@ -98,11 +98,12 @@ We support [ViT](https://arxiv.org/abs/2010.11929) and [DEIT](https://arxiv.org/abs/2012.12877) implementations. ViT models trained under the DEIT settings: -model | resolution | Top-1 | Top-5 | ---------- | :--------: | ----: | ----: | -ViT-s16 | 224x224 | 79.4 | 94.7 | -ViT-b16 | 224x224 | 81.8 | 95.8 | -ViT-l16 | 224x224 | 82.2 | 95.8 | +model | resolution | Top-1 | Top-5 | Download | +--------- | :--------: | ----: | ----: | :-------: | +ViT-ti16 | 224x224 | 73.4 | 91.9 | [ckpt](https://storage.googleapis.com/tf_model_garden/vision/vit/vit-deit-imagenet-ti16.tar.gz) | +ViT-s16 | 224x224 | 79.4 | 94.7 | [ckpt](https://storage.googleapis.com/tf_model_garden/vision/vit/vit-deit-imagenet-s16.tar.gz) | +ViT-b16 | 224x224 | 81.8 | 95.8 | [ckpt](https://storage.googleapis.com/tf_model_garden/vision/vit/vit-deit-imagenet-b16.tar.gz) | +ViT-l16 | 224x224 | 82.2 | 95.8 | [ckpt](https://storage.googleapis.com/tf_model_garden/vision/vit/vit-deit-imagenet-l16.tar.gz) | @@ -160,9 +161,9 @@ training features including: | Backbone | Resolution | Epochs | FLOPs (B) | Params (M) | Box AP | Download | | ------------ |:-------------:| -------:|--------------:|-----------:|--------:|---------:| -| SpineNet-49 | 640x640 | 500 | 85.4| 28.5 | 44.2 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/retinanet/coco_spinenet49_tpu.yaml) \| [TB.dev](https://tensorboard.dev/experiment/n2UN83TkTdyKZn3slCWulg/#scalars&_smoothingWeight=0)| -| SpineNet-96 | 1024x1024 | 500 | 265.4 | 43.0 | 48.5 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/retinanet/coco_spinenet96_tpu.yaml) \| [TB.dev](https://tensorboard.dev/experiment/n2UN83TkTdyKZn3slCWulg/#scalars&_smoothingWeight=0)| -| SpineNet-143 | 1280x1280 | 500 | 524.0 | 67.0 | 50.0 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/retinanet/coco_spinenet143_tpu.yaml) \| [TB.dev](https://tensorboard.dev/experiment/n2UN83TkTdyKZn3slCWulg/#scalars&_smoothingWeight=0)| +| SpineNet-49 | 640x640 | 500 | 85.4| 28.5 | 44.2 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/retinanet/coco_spinenet49_tpu.yaml) \| [ckpt](https://storage.googleapis.com/tf_model_garden/vision/spinenet/spinenet-49-i640.tar.gz) \| [TB.dev](https://tensorboard.dev/experiment/n2UN83TkTdyKZn3slCWulg/#scalars&_smoothingWeight=0)| +| SpineNet-96 | 1024x1024 | 500 | 265.4 | 43.0 | 48.5 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/retinanet/coco_spinenet96_tpu.yaml) \| [ckpt](https://storage.googleapis.com/tf_model_garden/vision/spinenet/spinenet-96-i1024.tar.gz) \| [TB.dev](https://tensorboard.dev/experiment/n2UN83TkTdyKZn3slCWulg/#scalars&_smoothingWeight=0)| +| SpineNet-143 | 1280x1280 | 500 | 524.0 | 67.0 | 50.0 | [config](https://github.com/tensorflow/models/blob/master/official/vision/configs/experiments/retinanet/coco_spinenet143_tpu.yaml) \| [ckpt](https://storage.googleapis.com/tf_model_garden/vision/spinenet/spinenet-143-i1280.tar.gz) \| [TB.dev](https://tensorboard.dev/experiment/n2UN83TkTdyKZn3slCWulg/#scalars&_smoothingWeight=0)| @@ -218,7 +219,7 @@ training features including: | Model | Backbone | Resolution | Steps | mIoU | Download | | ---------- | :----------------: | :--------: | ----: | ---: | --------:| | DeepLabV3 | Dilated Resnet-101 | 512x512 | 30k | 78.7 | | -| DeepLabV3+ | Dilated Resnet-101 | 512x512 | 30k | 79.2 | | +| DeepLabV3+ | Dilated Resnet-101 | 512x512 | 30k | 79.2 | [ckpt](https://storage.googleapis.com/tf_model_garden/vision/deeplabv3plus/dilated-resnet-101-deeplabv3plus.tar.gz) | -- GitLab