[DEV] ssds.pytorch v1.5 roadmap

[foreverYoungGitHub]

This is the roadmap for ssds.pytorch v1.5. The development for ssds.pytorch v1.5 is fully reconstruct and almost done. The main features are listed at here:

Documentations:

• [x] install;
• [x] basic usage;
• [x] basic parameters;
• [ ] sample tutorials;
• [x] basic python api.

Framework:

• [x] add training, validation, and inference support for pytorch v1.5 based model;
• [x] add support to convert the model from pytorch to onnx, and allows the user to convert the onnx model the tensorrt 7 through the code in retinanet-examples.

Dataset:

• [x] remove the current the voc dataset;
• [x] add the dalicoco and dalitfrecord dataset for fast data loading.

Anchor box matching:

• [x] remove the current anchor mathcing strategy;
• [x] add the anchor box matching for each level to make user understand ssd-like training and inference easier.

Loss:

• [x] FocalLoss
• [x] SmoothL1
• [x] IOU, GIOU, DIOU, CIOU Loss
• [x] MultiBoxLoss (Not recommend, not fully tested)

Pipeline:

• [x] add DataParallel for basic multiple gpu or single gpu training (slow)
• [x] add apex for multiple gpu training (fast)

Visualization:

• [x] add visualization for anchor strategy in each feature map (the distribution of scale and ratio in the dataset);
• [x] add visualization for defualt anchor boxes in each feature map;
• [ ] prepare the images for readme.

Support SSDs head:

• [x] ssd;
• [x] fpn in retinanet;
• [x] bifpn in efficientdet;
• [x] yolov3 and yolov4
• [x] shelf in shelfnet

Support backbone (feature extractor):

• [x] resnet
• [x] regnetx
• [x] mobilenet v1 and v2
• [x] shufflenet v2
• [x] darknet
• [x] densenet
• [x] efficientNet (memory cost)

Others:

• [x] Provide the dockerfile to allow user directly build the ssds.pytorch docker quickly;
• [x] Provide the setup.py to allow user directly install the ssds.pytorch by pip;
• [ ] Prepare the pretrained models for different backbone and detection heads.

Bug Fix:

Please let me know if you have any problem when you use the ssds.pytorch or any suggestion to make the ssds.pytorch better!