tdolan21/detr-resnet-50-api

This project is an API with a demo application utilizing facebook's detr-resnet-50 model for object detection in both image and video. The usage in video is experimental as the original model is trained for images.

Object Detection API using DETR

This project is an object detection API built using FastAPI and Streamlit. It utilizes the DETR (DEtection TRansformer) model for end-to-end object detection.

The detr-resnet-50 model is trained on the COCO image dataset.

Usage

• Clone the repository to your machine:

git clone https://github.com/tdolan21/detr-resnet-50-api.git
cd detr-resnet-50-api
pip install -r requirements.txt

• Windows machines can run the code by using the batch script:

 ./run.bat

• Linux machines can use the bash script:

chmod +x run.sh
./run.sh

Once you have run the startup script the application will be available at:

localhost:8000
localhost:8501

Key Features:

• Object Detection in Images: Detect and label objects in uploaded images.
• Object Detection in Videos: Detect, label, and annotate objects frame by frame in uploaded videos.
• Fast, Accurate & Scalable: Utilizes the DETR model for precise object detection.
• Interactive: Offers colorful visualization for detected objects.

Endpoints:

1. / (GET)

   • Welcomes users and provides a brief introduction to the API.

2. /detect/ (POST)

   • Accepts an image file.
   • Detects and labels objects in the image.
   • Returns a list of detected objects with their labels, confidence scores, and bounding box coordinates.

3. /detect_video/ (POST)

   • Accepts a video file.
   • Detects and labels objects frame by frame in the video.
   • Provides an output video with annotated objects.
   • Returns a URL for the annotated video download and a list of detections for each frame.

Usage:

Simply upload an image or video through the respective endpoint and receive annotated results with detected objects.

Citation

@article{DBLP:journals/corr/abs-2005-12872,
  author    = {Nicolas Carion and
               Francisco Massa and
               Gabriel Synnaeve and
               Nicolas Usunier and
               Alexander Kirillov and
               Sergey Zagoruyko},
  title     = {End-to-End Object Detection with Transformers},
  journal   = {CoRR},
  volume    = {abs/2005.12872},
  year      = {2020},
  url       = {https://arxiv.org/abs/2005.12872},
  archivePrefix = {arXiv},
  eprint    = {2005.12872},
  timestamp = {Thu, 28 May 2020 17:38:09 +0200},
  biburl    = {https://dblp.org/rec/journals/corr/abs-2005-12872.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}