Quickvision

This new release cleans up our codebases internally. This makes the library more robust and easier to maintain.
Also, we have new examples and prototype APIs.

Code Cleanup and Bug Fixes

- We cleaned up code internally and created a stable `requirements.txt` file. Now, we pin versions in our requirements instead of hardcoding them in CI checks.
- It is recommended to install PyTorch as mentioned on PyTorch website and NOT run `pip install -r requirements.txt`.
- Our `requirements.txt` file helps us to view Dependabot alerts as well as see sub dependencies graph.

56 Cleaning up Type Annotations
60 Fixed some typos and small fixes.
63 Drop timm requirements.
65 Clean up of codebase like Pl_bolts, making it easier to create a model zoo
77 Fixing CI and file encoding issue over MacOS and windows.
69 Enhanced Dummy Detection dataset for better internal testing
79 Refactored tests to check for normalized bounding boxes and class starts.
88 Added `requirements.txt` file which makes it easier to keep consistency.
90 Fixed Detr Name and weights.

New Models and Tutorials

76 Supports Wide Resnet for Classification as well as FPN based backbone in Detection.
81 Example to train models with Weights and Biases.
74 Example to train DETR over COCO Dataset.
68 SWAV Weights for Resnet models in Detection and Classification tasks.

Prototype APIs

As a small release, we have started prototyping new APIs. We expect them to be stable in subsequent releases.

86 Losses API for common losses in Computer Vision tasks.
80 Layers API for layers pertaining to Computer Vision.


Documentation

- We are planning to build Sphinx Documentation with autodoc features. Thanks to SauravMaheshkar .
- This is work in progress and contributions are welcome

54 Adds Sphinx Docs
67 Fixes a Few issues with Sphinx
58 Added CI to deploy Docs over GitHub pages.

Since this release is fully compatible with PyTorch Lightning 1.1, people can use all features of Lightning such as Shared Training, etc.

Super thanks to hassiahk zhiqwang zlapp SauravMaheshkar for making this release possible!

In this initial open sourcing, we have provided training APIs for detection as well as classification tasks.
- Supported some torchvision CNNs training.
- Supported Detection models from torchvision Faster RCNN 7 and Retina Net 6 .
- Supported DETR (Detection Transformers) for Object detection transfer learning through `torch.hub` 21
- Supports Some backbones which can be trained through CNN Trainer.
- Supported Pretrained weights other than imagenet for few models 27 .
- Have Sanity Fir APIs for all the above models 46 51 .
- PyTorch Lightning trainers for all above.
- Added lot of tutorials and notebooks for users to learn 52 . We hope you get the feel of library and use it.

Thanks to our awesome contributors hassiahk ramaneswaran for helping in this release.

- Note that this current release requires PyTorch 1.7 and torchvision 0.8.1.

FAQs about Quickvision: -
- Will this support fastai, ignite, catalyst, TensorFlow, Keras, etc ?

No, it will not, it is only based on torchvision and PyTorch Lightning (optionally).

- Does this have internal data representation and complex classes ?

No, it only abstracts code and avoids use of mixins, multiple inheritances and such confusions.
We deal only in `Tensors` and hence it is much easier to use.

- Does it provides Augmentations ?

We avoid binding to any augmentations and leave that flexibility to user. You may use any augmentation library such as Torchvision, Albumentations, etc. This is left entirely to end users, as augmentations heavily depend on datasets and user's choice of libraries.

- Is Quickvision a Framework ?

No ! This is mere extension to torchvision. We do not wish to be a framework.
A framework is end to end library that would do Data loading, Preprocessing, Model Creation, Training, Post Processing and Visualization.
Quickvision only accelerates the the Model Creation and Training parts.
We feel that rest part are better left to end user for wiser decisions.

- Is Quickvision limited to Object Detection ?

No ! It is a Computer Vision library, which would later extend to other tasks as well.
Right now we support Image Classification and Object Detection.

- Quickvision is very similar to torchvision then why is it there ?

Torchvision is great, in fact quickvision is created taking heavy inspirations and ideas from torchvision, we have tried to keep API similar. In a way Quickvision extends torchvision, we would love to host other computer vision tasks, more models and their implementations, provide a clean training API. It does not intend to substitute torchvision instead use it as base.