EfficientDet-Object-Detection

PyTorch Implementation of EfficientDet: Scalable and Efficient Object Detection EfficientDet: Scalable and Efficient Object Detection paper

To run object Detection on Google Colab:

https://colab.research.google.com/drive/1dugPLJDZJLFhya9pcdU-gk1d7YrzZJQp?usp=sharing

NOTE: Upload your video in the "Test_Videos" folder and the detected output video will be saved in "output_folder" you can download it later

Create a data folder under the repository,

cd {repo_root}
mkdir data

• COCO: Download the coco images and annotations from coco website. Make sure to put the files as the following structure:

  COCO
  ├── annotations
  │   ├── instances_train2017.json
  │   └── instances_val2017.json
  │── images
      ├── train2017
      └── val2017
  

• Train your model by running python train.py
• Evaluate mAP for COCO dataset by running python mAP_evaluation.py
• Test your model for COCO dataset by running python test_dataset.py --pretrained_model path/to/trained_model
• Test your model for video by running python test_video.py --pretrained_model path/to/trained_model --input path/to/input/file --output path/to/output/file