Object Detection

RetiaNet

  • Object detection model RetinaNet has been formed by making two improvements over existing single stage object detection models - Feature Pyramid Networks and Focal Loss.

image

Feature Pyramid Network

  • Feature image pyramids used to detect objects with varying scales in an image.

  • In feature image pyramids, we take an input image and subsample it into lower resolution and smaller size images

  • With the advancements of deep learning, we now use the pyramidal hierarchical structure with CNNs.

  • In a CNN architecture, the output size of feature maps decreases after each successive block of convolutional operations, and forms a pyramidal structure.

Focal Loss

  • Focal Loss is an enhancement over Cross-Entropy Loss and is introduced to handle the class imbalance problem with single-stage object detection models.

  • Single Stage models suffer from a extreme foreground-background class imbalance problem due to dense sampling of anchor boxes

  • Focal Loss reduces the loss contribution from easy examples and increases the importance of correcting misclassified examples.

Advantages

  • Highest accuracy object detectors to date are based on a two-stage approach popularized by R-CNN

  • Surpassing the accuracy of all existing state-of-the-art two-stage detector

UNETs

UNet is a fully convolutional network (FCN) used for image segmentation. The goal is to predict each pixel’s class in an image.

Architecture

  • Three Main Components):

    • Encoder or Downsampling Path

    • Bottleneck

    • Decoder or Upsampling Path

Downsampling Path:

  • Consists of two convolution layers each followed by a ReLU activation function and a 2x2 max pooling operation for downsampling.

  • At each downsampling step we double the number of feature channels

Bottleneck:

  • part of the network is between the contracting and expanding paths. The bottleneck is built from 2 convolutional layers (with batch normalization) and with dropout.

Upsampling Path:

  • Every step in the decoder path consists of an upsampling of the feature map followed by a 2x2 convolution, a concatenation with the corresponding feature map from the downsampling path, and two convolutions layers, each followed by a ReLU.

Final Layer:

  • A 1x1 convolution is used to map each feature vector to the desired number of classes.

Loss function:

  • The energy function is computed by a pixel-wise softmax over the final feature map and then applied cross-entropy loss function.

Metrics

  • IoU (intersection of union) -> Area overlap / Area of Union

Notes

  • Means converting a high resolution image to a low resolution image. By down sampling, the model better understands What is present in the image, but it loses the information of Where it is present.

Advantages

  • UNet combines the location information from the downsampling path to finally obtain a general information combining localisation and context, which is necessary to predict a good segmentation map.