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