DETR

DETR object detection model formulates the problem of object detection as a direct set prediction problem effectively removing the need for hand designed components like a non max supression or anchor box generation that explicitly encode the knowledge of the task. It uses a transformer encoder-decoder based architecture to output set predictions in parallel.

Transformer architecture

The transformer architecture is slightly different from Vaswani et al and uses a positional encoding at each encoder and decoder layer.

Loss Function

The loss functions used are a giou loss for bounding box, a negative log likelihood loss for class predictions and an L1 loss for the predicted and ground truth boxes.

Dataset

The DETR model is being trained on the Pascal VOC 2012 object detection dataset with 20 classes and around 5000 train and validation images.

sample images

2008_000142	2008_000196

sample annotation file:

<annotation>
	<folder>VOC2012</folder>
	<filename>2008_000041.jpg</filename>
	<source>
		<database>The VOC2008 Database</database>
		<annotation>PASCAL VOC2008</annotation>
		<image>flickr</image>
	</source>
	<size>
		<width>500</width>
		<height>375</height>
		<depth>3</depth>
	</size>
	<segmented>0</segmented>
	<object>
		<name>pottedplant</name>
		<pose>Unspecified</pose>
		<truncated>0</truncated>
		<occluded>0</occluded>
		<bndbox>
			<xmin>54</xmin>
			<ymin>189</ymin>
			<xmax>84</xmax>
			<ymax>214</ymax>
		</bndbox>
		<difficult>1</difficult>
	</object>
	<object>
		<name>person</name>
		<pose>Rear</pose>
		<truncated>1</truncated>
		<occluded>0</occluded>
		<bndbox>
			<xmin>358</xmin>
			<ymin>188</ymin>
			<xmax>469</xmax>
			<ymax>359</ymax>
		</bndbox>
		<difficult>0</difficult>
	</object>
</annotation>

---

Project Organization

├── LICENSE
├── Makefile           <- Makefile with commands like `make data` or `make train`
├── README.md          <- The top-level README for developers using this project.
├── data
│   ├── external       <- Data from third party sources.
│   ├── interim        <- Intermediate data that has been transformed.
│   ├── processed      <- The final, canonical data sets for modeling.
│   └── raw            <- The original, immutable data dump.
│
├── docs               <- A default Sphinx project; see sphinx-doc.org for details
│
├── models             <- Trained and serialized models, model predictions, or model summaries
│
├── notebooks          <- Jupyter notebooks. Naming convention is a number (for ordering),
│                         the creator's initials, and a short `-` delimited description, e.g.
│                         `1.0-jqp-initial-data-exploration`.
│
├── references         <- Data dictionaries, manuals, and all other explanatory materials.
│
├── reports            <- Generated analysis as HTML, PDF, LaTeX, etc.
│   └── figures        <- Generated graphics and figures to be used in reporting
│
├── requirements.txt   <- The requirements file for reproducing the analysis environment, e.g.
│                         generated with `pip freeze > requirements.txt`
│
├── setup.py           <- makes project pip installable (pip install -e .) so src can be imported
├── src                <- Source code for use in this project.
│   ├── __init__.py    <- Makes src a Python module
│   │
│   ├── data           <- Scripts to download or generate data
│   │   └── make_dataset.py
│   │
│   ├── features       <- Scripts to turn raw data into features for modeling
│   │   └── build_features.py
│   │
│   ├── models         <- Scripts to train models and then use trained models to make
│   │   │                 predictions
│   │   ├── predict_model.py
│   │   └── train_model.py
│   │
│   └── visualization  <- Scripts to create exploratory and results oriented visualizations
│       └── visualize.py
│
└── tox.ini            <- tox file with settings for running tox; see tox.readthedocs.io

---

Project based on the cookiecutter data science project template. #cookiecutterdatascience

"# DETR"