FasterRCNN for object detection. 
/home/user/jupyter/env/lib/python3.7/site-packages/torch/cuda/__init__.py:52: UserWarning: CUDA initialization: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx (Triggered internally at  /pytorch/c10/cuda/CUDAFunctions.cpp:100.)
  return torch._C._cuda_getDeviceCount() > 0

FasterRCNN paper - Implementation

Functions to create the model

get_fasterrcnn_model[source]

get_fasterrcnn_model(arch_str, num_classes, pretrained=True, pretrained_backbone=True, trainable_layers=5, min_size=800, max_size=1333, image_mean=None, image_std=None, rpn_anchor_generator=None, rpn_head=None, rpn_pre_nms_top_n_train=2000, rpn_pre_nms_top_n_test=1000, rpn_post_nms_top_n_train=2000, rpn_post_nms_top_n_test=1000, rpn_nms_thresh=0.7, rpn_fg_iou_thresh=0.7, rpn_bg_iou_thresh=0.3, rpn_batch_size_per_image=256, rpn_positive_fraction=0.5, box_roi_pool=None, box_head=None, box_predictor=None, box_score_thresh=0.05, box_nms_thresh=0.5, box_detections_per_img=100, box_fg_iou_thresh=0.5, box_bg_iou_thresh=0.5, box_batch_size_per_image=512, box_positive_fraction=0.25, bbox_reg_weights=None)

Creates FasterRCNN model with resnet backbone

get_fasterrcnn_model_swin[source]

get_fasterrcnn_model_swin(arch_str, num_classes, pretrained=False, pretrained_backbone=True, min_size=800, max_size=1333, image_mean=None, image_std=None, rpn_anchor_generator=None, rpn_head=None, rpn_pre_nms_top_n_train=2000, rpn_pre_nms_top_n_test=1000, rpn_post_nms_top_n_train=2000, rpn_post_nms_top_n_test=1000, rpn_nms_thresh=0.7, rpn_fg_iou_thresh=0.7, rpn_bg_iou_thresh=0.3, rpn_batch_size_per_image=256, rpn_positive_fraction=0.5, box_roi_pool=None, box_head=None, box_predictor=None, box_score_thresh=0.05, box_nms_thresh=0.5, box_detections_per_img=100, box_fg_iou_thresh=0.5, box_bg_iou_thresh=0.5, box_batch_size_per_image=512, box_positive_fraction=0.25, bbox_reg_weights=None)

Creates FasterRCNN model with swin transformer backbone

Custom FasterRCNN

To create a model, which you can pass to a ObjDetLearner simply create a partial with the functions get_fasterrcnn_model (resnet backbone) or get_fasterrcnn_model_swin (swin transformer backbone).

from functools import partial

custom_fasterrcnn = partial(get_fasterrcnn_model, arch_str="resnet18", 
                            pretrained=True, pretrained_backbone=True, 
                            min_size=600, max_size=600)

When building the Learner, the number of classes are getting passed to this partial function and the model is ready for training.

custom_fasterrcnn(num_classes=5)

No pretrained coco model found for fasterrcnn_resnet18
This does not affect the backbone.

FasterRCNN(
  (transform): GeneralizedRCNNTransform(
      Normalize(mean=[0.0, 0.0, 0.0], std=[1.0, 1.0, 1.0])
      Resize(min_size=(600,), max_size=600, mode='bilinear')
  )
  (backbone): BackboneWithFPN(
    (body): IntermediateLayerGetter(
      (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
      (bn1): FrozenBatchNorm2d(64)
      (relu): ReLU(inplace=True)
      (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
      (layer1): Sequential(
        (0): BasicBlock(
          (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64)
          (relu): ReLU(inplace=True)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64)
        )
        (1): BasicBlock(
          (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(64)
          (relu): ReLU(inplace=True)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(64)
        )
      )
      (layer2): Sequential(
        (0): BasicBlock(
          (conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128)
          (relu): ReLU(inplace=True)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128)
          (downsample): Sequential(
            (0): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(128)
          )
        )
        (1): BasicBlock(
          (conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(128)
          (relu): ReLU(inplace=True)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(128)
        )
      )
      (layer3): Sequential(
        (0): BasicBlock(
          (conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256)
          (relu): ReLU(inplace=True)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256)
          (downsample): Sequential(
            (0): Conv2d(128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(256)
          )
        )
        (1): BasicBlock(
          (conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(256)
          (relu): ReLU(inplace=True)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(256)
        )
      )
      (layer4): Sequential(
        (0): BasicBlock(
          (conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512)
          (relu): ReLU(inplace=True)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512)
          (downsample): Sequential(
            (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)
            (1): FrozenBatchNorm2d(512)
          )
        )
        (1): BasicBlock(
          (conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn1): FrozenBatchNorm2d(512)
          (relu): ReLU(inplace=True)
          (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): FrozenBatchNorm2d(512)
        )
      )
    )
    (fpn): FeaturePyramidNetwork(
      (inner_blocks): ModuleList(
        (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1))
        (1): Conv2d(128, 256, kernel_size=(1, 1), stride=(1, 1))
        (2): Conv2d(256, 256, kernel_size=(1, 1), stride=(1, 1))
        (3): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))
      )
      (layer_blocks): ModuleList(
        (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
        (1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
        (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
        (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      )
      (extra_blocks): LastLevelMaxPool()
    )
  )
  (rpn): RegionProposalNetwork(
    (anchor_generator): AnchorGenerator()
    (head): RPNHead(
      (conv): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (cls_logits): Conv2d(256, 3, kernel_size=(1, 1), stride=(1, 1))
      (bbox_pred): Conv2d(256, 12, kernel_size=(1, 1), stride=(1, 1))
    )
  )
  (roi_heads): RoIHeads(
    (box_roi_pool): MultiScaleRoIAlign()
    (box_head): TwoMLPHead(
      (fc6): Linear(in_features=12544, out_features=1024, bias=True)
      (fc7): Linear(in_features=1024, out_features=1024, bias=True)
    )
    (box_predictor): FastRCNNPredictor(
      (cls_score): Linear(in_features=1024, out_features=5, bias=True)
      (bbox_pred): Linear(in_features=1024, out_features=20, bias=True)
    )
  )
)

Prebuilt models

There are some prebuilt model partials, which you can use instantly:

Resnet Backbone:

Swin Transformer Backbone: