Model Overview

With a ResNet-50 backbone and a number of architectural modifications, this version provides better accuracy and performance.

Model Architecture

Our implementation is based on the existing model from the TensorFlow models repository. The network was altered in order to improve accuracy and increase throughput. Changes include:

• Replacing the VGG backbone with the more popular ResNet50.
• Adding multi-scale detection to the backbone using Feature Pyramid Networks.
• Replacing the original hard negative mining loss function with Focal Loss.
• Decreasing the input size to 320 x 320.

Training

This model was trained using script available on NGC and in GitHub repo

Dataset

The following datasets were used to train this model:

• COCO 2017 - Dataset for large-scale object detection, segmentation and captioning.

Performance

Performance numbers for this model are available in NGC

References

• original paper
• NVIDIA model implementation in NGC
• NVIDIA model implementation on GitHub

License

This model was trained using open-source software available in Deep Learning Examples repository. For terms of use, please refer to the license of the script and the datasets the model was derived from.