/FasterRCNN3

A Detection Toolbox for Tensorflow2

Primary LanguageJupyter NotebookApache License 2.0Apache-2.0

TFDetection

TFDetection(tfdet) is a detection toolbox based on Tensorflow2 and Keras.

Overview of Components

Detector
Object Detection Segmentation Anomaly Detection
  • Faster R-CNN(2015)
  • RetinaNet(2017)
  • YoloV3(2018)
  • YoloV3 Tiny(2018)
  • Cascade R-CNN(2018)
  • FCOS(2019)
  • Hybrid Task Cascade(2019)
  • EfficientDet(2019)
  • YoloV4(2020)
  • YoloV4 Tiny(2020)
  • EfficientDet Lite(2020)
  • Semantic Segmentation
    • FCN(2015)
    • UNet(2016)
    • PSPNet(2017)
    • DeepLab V3(2017)
    • DeepLab V3+(2018)
    • UNet++(2018)
    • UperNet(2018)
  • Instance Segmentation
    • Mask R-CNN(2017)
    • Cascade R-CNN(2018)
    • Hybrid Task Cascade(2019)
  • SPADE(2020)
  • PaDiM(2020)
  • PatchCore(2021)
Components
Backbone Neck Other
  • VGGNet(2015)
  • ResNet(2016)
  • Wide ResNet(2016)
  • DenseNet(2016)
  • ResNeXt(2016)
  • MobileNet(2017)
  • MobileNetV2(2018)
  • DarkNet(2018)
  • MobileNetV3(2019)
  • EfficientNet(2019)
  • EfficientNet Lite(2020)
  • ResNeSt(2020)
  • CSP DarkNet(2020)
  • SwinTransformer(2021)
  • EfficientNetV2(2021)
  • SwinTransformerV2(2022)
  • FPN(2017)
  • PANet(2018)
  • BiFPN(2019)
  • Assign
    • Max Iou(2015)
    • Center Region(2019)
    • Point(2019)
    • Adaptive Training Sample Selection(2020)
  • Metric
    • Mean Average Precision

Requirements

  • Python 3.8▲
  • tensorflow 2.4▲ (If you installed tensorflow latest, more backbone is available.)
  • opencv-python
  • albumentations

Contributor