An open API service indexing awesome lists of open source software.

https://github.com/ternaus/check_orientation

Model to check if image was rotated by 90, 180, 270 degrees.
https://github.com/ternaus/check_orientation

computer-vision deep-learning image-classification orientation-detection

Last synced: 12 months ago
JSON representation

Model to check if image was rotated by 90, 180, 270 degrees.

Awesome Lists containing this project

README

          

# Check orientation

![](https://habrastorage.org/webt/hb/ru/ii/hbruiiuortx05lxfuokzw1skheu.jpeg)

Models to check if image was rotated by 0, 90, 180, 270 degrees.

## Installation
`pip install -U check_orientation`

### Example inference

Colab notebook with the example: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1HIGM_b8rH20N414ROZB0HN9w5p4ko2Bd?usp=sharing)

## Training

### Define the config.
Example at [check_orientation/configs](check_orientation/configs)

### Define the environmental variable `TRAIN_IMAGE_PATH` that points to the folder with train dataset.
Example:
```bash
export TRAIN_IMAGE_PATH=
```

### Define the environmental variable `VAL_IMAGE_PATH` that points to the folder with validation dataset.
Example:
```bash
export VAL_IMAGE_PATH=
```

### Training
```
python -m check_orientation.train -c
```

### Inference

```bash
python -m torch.distributed.launch --nproc_per_node= check_orientation/inference.py \
-i \
-c \
-w \
-o \
--fp16
```

### Pre-trained models
Models were pre-trained on the [OpenImages dataset](https://storage.googleapis.com/openimages/web/index.html).

| Models | Validation accuracy | Config file | Weights |
| ------------- |:--------------------:| :------------:| :------: |
| swsl_resnext50_32x4d|0.9128| [Link](check_orientation/configs/2020-11-16.yaml)| [Link](https://github.com/ternaus/check_orientation/releases/download/v0.0.3/2020-11-16_resnext50_32x4d.zip)|