models mmd 3x mask rcnn_swin t p4 w7_fpn_1x_coco - Azure/azureml-assets GitHub Wiki

mmd-3x-mask-rcnn_swin-t-p4-w7_fpn_1x_coco

Overview

mask-rcnn_swin-t-p4-w7_fpn_1x_coco model is from OpenMMLab's MMDetection library. This paper presents a new vision Transformer, called Swin Transformer, that capably serves as a general-purpose backbone for computer vision. Challenges in adapting Transformer from language to vision arise from differences between the two domains, such as large variations in the scale of visual entities and the high resolution of pixels in images compared to words in text. To address these differences, we propose a hierarchical Transformer whose representation is computed with Shifted windows. The shifted windowing scheme brings greater efficiency by limiting self-attention computation to non-overlapping local windows while also allowing for cross-window connection. This hierarchical architecture has the flexibility to model at various scales and has linear computational complexity with respect to image size. These qualities of Swin Transformer make it compatible with a broad range of vision tasks, including image classification (87.3 top-1 accuracy on ImageNet-1K) and dense prediction tasks such as object detection (58.7 box AP and 51.1 mask AP on COCO test-dev) and semantic segmentation (53.5 mIoU on ADE20K val). Its performance surpasses the previous state-of-the-art by a large margin of +2.7 box AP and +2.6 mask AP on COCO, and +3.2 mIoU on ADE20K, demonstrating the potential of Transformer-based models as vision backbones. The hierarchical design and the shifted window approach also prove beneficial for all-MLP architectures.

Training Details

Training Data

The model developers used COCO dataset for training the model.

Training Procedure

Training Techniques:

  • AdamW

Training Memory (GB): 7.6

Epochs: 12

Training Resources: 8x V100 GPUs

Evaluation Results

mask AP: 39.3

License

apache-2.0

Inference Samples

Inference type Python sample (Notebook) CLI with YAML
Real time image-instance-segmentation-online-endpoint.ipynb image-instance-segmentation-online-endpoint.sh
Batch image-instance-segmentation-batch-endpoint.ipynb image-instance-segmentation-batch-endpoint.sh

Finetuning Samples

Task Use case Dataset Python sample (Notebook) CLI with YAML
Image instance segmentation Image instance segmentation fridgeObjects fridgeobjects-instance-segmentation.ipynb fridgeobjects-instance-segmentation.sh

Evaluation Samples

Task Use case Dataset Python sample (Notebook)
Image instance segmentation Image instance segmentation fridgeObjects image-instance-segmentation.ipynb

Sample input and output

Sample input

{
  "input_data": {
    "columns": [
      "image"
    ],
    "index": [0, 1],
    "data": ["image1", "image2"]
  }
}

Note: "image1" and "image2" string should be in base64 format or publicly accessible urls.

Sample output

[
    {
        "boxes": [
            {
                "box": {
                    "topX": 0.1,
                    "topY": 0.2,
                    "bottomX": 0.8,
                    "bottomY": 0.7
                },
                "label": "carton",
                "score": 0.98,
                "polygon": [
                    [ 0.576, 0.680,  …]
                ]
            }
        ]
    },
    {
        "boxes": [
            {
                "box": {
                    "topX": 0.2,
                    "topY": 0.3,
                    "bottomX": 0.6,
                    "bottomY": 0.5
                },
                "label": "can",
                "score": 0.97,
                "polygon": [
                    [ 0.58, 0.7,  …]
                ]
            }
        ]
    }
]

Note: Please refer to instance segmentation output data schema for more detail.

Visualization of inference result for a sample image

is visualization

Version: 15

Tags

SharedComputeCapacityEnabled openmmlab_model_id : mmd-3x-mask-rcnn_swin-t-p4-w7_fpn_1x_coco training_dataset : COCO license : apache-2.0 model_specific_defaults : ordereddict({'apply_deepspeed': 'false', 'apply_ort': 'false'}) task : image-segmentation hiddenlayerscanned inference_compute_allow_list : ['Standard_DS3_v2', 'Standard_D4a_v4', 'Standard_D4as_v4', 'Standard_DS4_v2', 'Standard_D8a_v4', 'Standard_D8as_v4', 'Standard_DS5_v2', 'Standard_D16a_v4', 'Standard_D16as_v4', 'Standard_D32a_v4', 'Standard_D32as_v4', 'Standard_D48a_v4', 'Standard_D48as_v4', 'Standard_D64a_v4', 'Standard_D64as_v4', 'Standard_D96a_v4', 'Standard_D96as_v4', 'Standard_FX4mds', 'Standard_F8s_v2', 'Standard_FX12mds', 'Standard_F16s_v2', 'Standard_F32s_v2', 'Standard_F48s_v2', 'Standard_F64s_v2', 'Standard_F72s_v2', 'Standard_FX24mds', 'Standard_FX36mds', 'Standard_FX48mds', 'Standard_E4s_v3', 'Standard_E8s_v3', 'Standard_E16s_v3', 'Standard_E32s_v3', 'Standard_E48s_v3', 'Standard_E64s_v3', 'Standard_NC4as_T4_v3', 'Standard_NC6s_v3', 'Standard_NC8as_T4_v3', 'Standard_NC12s_v3', 'Standard_NC16as_T4_v3', 'Standard_NC24s_v3', 'Standard_NC64as_T4_v3', 'Standard_NC24ads_A100_v4', 'Standard_NC48ads_A100_v4', 'Standard_NC96ads_A100_v4', 'Standard_ND96asr_v4', 'Standard_ND96amsr_A100_v4', 'Standard_ND40rs_v2'] evaluation_compute_allow_list : ['Standard_NC4as_T4_v3', 'Standard_NC6s_v3', 'Standard_NC8as_T4_v3', 'Standard_NC12s_v3', 'Standard_NC16as_T4_v3', 'Standard_NC24s_v3', 'Standard_NC64as_T4_v3', 'Standard_NC96ads_A100_v4', 'Standard_ND96asr_v4', 'Standard_ND96amsr_A100_v4', 'Standard_ND40rs_v2'] finetune_compute_allow_list : ['Standard_NC4as_T4_v3', 'Standard_NC6s_v3', 'Standard_NC8as_T4_v3', 'Standard_NC12s_v3', 'Standard_NC16as_T4_v3', 'Standard_NC24s_v3', 'Standard_NC64as_T4_v3', 'Standard_NC96ads_A100_v4', 'Standard_ND96asr_v4', 'Standard_ND96amsr_A100_v4', 'Standard_ND40rs_v2']

View in Studio: https://ml.azure.com/registries/azureml/models/mmd-3x-mask-rcnn_swin-t-p4-w7_fpn_1x_coco/version/15

License: apache-2.0

Properties

SharedComputeCapacityEnabled: True

SHA: fff646d3dda72d8c794471bfaa75b4db0adb7610

finetuning-tasks: image-instance-segmentation

finetune-min-sku-spec: 4|1|28|176

finetune-recommended-sku: Standard_NC4as_T4_v3, Standard_NC6s_v3, Standard_NC8as_T4_v3, Standard_NC12s_v3, Standard_NC16as_T4_v3, Standard_NC24s_v3, Standard_NC64as_T4_v3, Standard_NC96ads_A100_v4, Standard_ND96asr_v4, Standard_ND96amsr_A100_v4, Standard_ND40rs_v2

evaluation-min-sku-spec: 4|1|28|176

evaluation-recommended-sku: Standard_NC4as_T4_v3, Standard_NC6s_v3, Standard_NC8as_T4_v3, Standard_NC12s_v3, Standard_NC16as_T4_v3, Standard_NC24s_v3, Standard_NC64as_T4_v3, Standard_NC96ads_A100_v4, Standard_ND96asr_v4, Standard_ND96amsr_A100_v4, Standard_ND40rs_v2

inference-min-sku-spec: 4|0|14|28

inference-recommended-sku: Standard_DS3_v2, Standard_D4a_v4, Standard_D4as_v4, Standard_DS4_v2, Standard_D8a_v4, Standard_D8as_v4, Standard_DS5_v2, Standard_D16a_v4, Standard_D16as_v4, Standard_D32a_v4, Standard_D32as_v4, Standard_D48a_v4, Standard_D48as_v4, Standard_D64a_v4, Standard_D64as_v4, Standard_D96a_v4, Standard_D96as_v4, Standard_FX4mds, Standard_F8s_v2, Standard_FX12mds, Standard_F16s_v2, Standard_F32s_v2, Standard_F48s_v2, Standard_F64s_v2, Standard_F72s_v2, Standard_FX24mds, Standard_FX36mds, Standard_FX48mds, Standard_E4s_v3, Standard_E8s_v3, Standard_E16s_v3, Standard_E32s_v3, Standard_E48s_v3, Standard_E64s_v3, Standard_NC4as_T4_v3, Standard_NC6s_v3, Standard_NC8as_T4_v3, Standard_NC12s_v3, Standard_NC16as_T4_v3, Standard_NC24s_v3, Standard_NC64as_T4_v3, Standard_NC24ads_A100_v4, Standard_NC48ads_A100_v4, Standard_NC96ads_A100_v4, Standard_ND96asr_v4, Standard_ND96amsr_A100_v4, Standard_ND40rs_v2

⚠️ **GitHub.com Fallback** ⚠️