fix markdown lint and some docs (#68)

* fix typos

* fix typos

* fix typos

* fix typos

.github/workflows/build.yml

@@ -1,6 +1,6 @@
 name: build
 
-on: [pull_request]
+on: [push, pull_request]
 
 jobs:
   lint:

README.md

@@ -15,7 +15,7 @@ Documentation: https://mmtracking.readthedocs.io/
 MMTracking is an open source video perception toolbox based on PyTorch.
 It is a part of the OpenMMLab project.
 
-The master branch works with PyTorch 1.3 to 1.6.
+The master branch works with PyTorch 1.3 to 1.7.
 
 <div align="left">
   <img src="https://user-images.githubusercontent.com/24663779/103343312-c724f480-4ac6-11eb-9c22-b56f1902584e.gif" width="800"/>
@@ -39,12 +39,10 @@ The master branch works with PyTorch 1.3 to 1.6.
 
   **Strong**: We reproduce state-of-the-art models and some of them even outperform the offical implementations.
 
-
 ## License
 
 This project is released under the [Apache 2.0 license](LICENSE).
 
-
 ## Changelog
 
 v0.5.0 was released in 04/01/2021.
@@ -78,7 +76,6 @@ Please refer to [install.md](docs/install.md) for install instructions.
 Please see [dataset.md](docs/dataset.md) and [quick_run.md](docs/quick_run.md) for the basic usage of MMTracking.
 We also provide usage [tutorials](docs/tutorials/).
 
-
 ## Contributing
 
 We appreciate all contributions to improve MMTracking. Please refer to [CONTRIBUTING.md](.github/CONTRIBUTING.md) for the contributing guideline.

configs/det/README.md

@@ -11,6 +11,7 @@ Please NOTE that there are some differences between the base config in MMTrackin
 1. `detector` is only a submodule of the `model`.
 
     For example, the config of Faster R-CNN in MMDetection follows
+
     ```python
     model = dict(
         type='FasterRCNN',
@@ -19,6 +20,7 @@ Please NOTE that there are some differences between the base config in MMTrackin
     ```
 
     But in MMTracking, the config follows
+
     ```python
     model = dict(
         detector=dict(
@@ -31,13 +33,15 @@ Please NOTE that there are some differences between the base config in MMTrackin
 2. `train_cfg` and `test_cfg` are merged into `model` / `detector`.
 
     In MMDetection, the configs follows
+
     ```python
     model = dict()
     train_cfg = dict()
     test_cfg = dict()
     ```
 
     While in MMTracking, the config follows
+
     ```python
     model = dict(
         detector=dict(

configs/mot/deepsort/README.md

@@ -29,7 +29,6 @@ The configs in this folder are basiclly for inference.
 Currently we do not support training ReID models.
 We directly use the ReID model from [Tracktor](https://github.com/phil-bergmann/tracking_wo_bnw). These missed features will be supported in the future.
 
-
 |    Detector     |  ReID  | Train Set | Test Set | Public | Inf time (fps) | MOTA | IDF1 | FP | FN | IDSw. | Config | Download |
 | :-------------: | :----: | :-------: | :------: | :----: | :------------: | :--: | :--: |:--:|:--:| :---: | :----: | :------: |
 | R50-FasterRCNN-FPN | -  | half-train | half-val | Y     | 28.3 |   46.0 | 46.6 | 289 | 82451 | 4581 | [config](sort_faster-rcnn_fpn_4e_mot17-public-half.py) |  [detector](https://download.openmmlab.com/mmtracking/v0.5/mot/faster-rcnn_r50_fpn_4e_mot17-half-64ee2ed4.pth) |

docs/changelog.md

@@ -1,10 +1,13 @@
 ## Changelog
 
 ### v0.5.0 (04/01/2021)
+
 #### Highlights
+
 - MMTracking is released!
 
 #### New Features
+
 - Support video object detection methods: [DFF](https://arxiv.org/abs/1611.07715), [FGFA](https://arxiv.org/abs/1703.10025), [SELSA](https://arxiv.org/abs/1907.06390)
 - Support multi object tracking methods: [SORT](https://arxiv.org/abs/1602.00763)/[DeepSORT](https://arxiv.org/abs/1703.07402), [Tracktor](https://arxiv.org/abs/1903.05625)
 - Support single object tracking methods: [SiameseRPN++](https://arxiv.org/abs/1812.11703)

docs/dataset.md

@@ -1,14 +1,13 @@
 ## Dataset Preparation
 
-
 This page provides the instructions for dataset preparation on existing benchmarks, include
 
 - Video Object Detection
-    - [ILSVRC](http://image-net.org/challenges/LSVRC/2017/)
+  - [ILSVRC](http://image-net.org/challenges/LSVRC/2017/)
 - Multiple Object Tracking
-    - [MOT Challenge](https://motchallenge.net/)
+  - [MOT Challenge](https://motchallenge.net/)
 - Single Object Tracking
-    - [LaSOT](http://vision.cs.stonybrook.edu/~lasot/)
+  - [LaSOT](http://vision.cs.stonybrook.edu/~lasot/)
 
 ### 1. Download Datasets
 
@@ -24,7 +23,6 @@ Notes:
 
 - For the training and testing of single object tracking task, the MSCOCO, ILSVRC and LaSOT datasets are needed.
 
-
 ```
 mmtracking
 ├── mmtrack
@@ -64,7 +62,6 @@ mmtracking
 |   |   ├── annotations
 ```
 
-
 ### 2. Convert Annotations
 
 We use [CocoVID](../mmtrack/datasets/parsers/coco_video_parser.py) to maintain all datasets in this codebase.

docs/install.md

@@ -115,10 +115,10 @@ conda activate open-mmlab
 conda install pytorch==1.6.0 torchvision==0.7.0 cudatoolkit=10.1 -c pytorch -y
 
 # install the latest mmcv
-pip install mmcv-full==latest+torch1.6.0+cu101 -f https://download.openmmlab.com/mmcv/dist/index.html
+pip install mmcv-full -f https://download.openmmlab.com/mmcv/dist/cu101/torch1.6.0/index.html
 
 # install mmdetection
-pip install git+https://github.com/open-mmlab/mmdetection.git
+pip install mmdet
 
 # install mmtracking
 git clone https://github.com/open-mmlab/mmtracking.git

docs/model_zoo.md

@@ -4,10 +4,8 @@
 
 - We use distributed training.
 - All pytorch-style pretrained backbones on ImageNet are from PyTorch model zoo.
-- For fair comparison with other codebases, we report the GPU memory as the maximum value of `torch.cuda.max_memory_allocated()` for all 8 GPUs.
-Note that this value is usually less than what `nvidia-smi` shows.
-- We report the inference time as the total time of network forwarding and post-processing, excluding the data loading time.
-Results are obtained with the script `tools/benchmark.py` which computes the average time on 2000 images.
+- For fair comparison with other codebases, we report the GPU memory as the maximum value of `torch.cuda.max_memory_allocated()` for all 8 GPUs. Note that this value is usually less than what `nvidia-smi` shows.
+- We report the inference time as the total time of network forwarding and post-processing, excluding the data loading time. Results are obtained with the script `tools/benchmark.py` which computes the average time on 2000 images.
 - Speed benchmark environments
 
     HardWare
@@ -24,23 +22,29 @@ Results are obtained with the script `tools/benchmark.py` which computes the ave
 ## Baselines of video object detection
 
 ### DFF
+
 Please refer to [DFF](../configs/vid/dff/README.md) for details.
 
 ### FGFA
+
 Please refer to [FGFA](../configs/vid/fgfa/README.md) for details.
 
 ### SELSA
+
 Please refer to [SELSA](../configs/vid/selsa/README.md) for details.
 
 ## Baselines of multiple object tracking
 
 ### SORT/DeepSORT
+
 Please refer to [SORT/DeepSORT](../configs/mot/deepsort/README.md) for details.
 
 ### Tracktor
+
 Please refer to [Tracktor](../configs/mot/tracktor/README.md) for details.
 
 ## Baselines of single object tracking
 
 ### SiameseRPN++
+
 Please refer to [SiameseRPN++](../configs/sot/siamese_rpn/README.md) for details.

docs/tutorials/customize_dataset.md

@@ -3,8 +3,8 @@
 To customize a new dataset, you can convert them to the existing CocoVID style or implement a totally new dataset.
 In MMTracking, we recommand to convert the data into CocoVID style and do the conversion offline, thus you can use the `CocoVideoDataset` directly. In this case, you only need to modify the config's data annotation pathes and the `classes`.
 
-
 ### Convert the dataset into CocoVID style
+
 #### The CocoVID annotation file
 
 The annotation json files in CocoVID style has the following necessary keys:
@@ -18,7 +18,6 @@ A simple example is presented at [here](../../tests/assets/demo_cocovid_data/ann
 
 The examples of converting existing datasets is presented at [here](../../tools/convert_datasets/).
 
-
 #### Modify the config
 
 After the data pre-processing, the users need to further modify the config files to use the dataset.
@@ -207,8 +206,6 @@ data = dict(
 
 ```
 
-
-
 ### Subset of existing datasets
 
 With existing dataset types, we can modify the class names of them to train subset of the annotations.

docs/tutorials/customize_mot_model.md

@@ -8,10 +8,8 @@ We basically categorize model components into 5 types.
 - reid: usually an independent ReID model to extract the feature embeddings from the cropped image, e.g., BaseReID.
 - track_head: the component to extract tracking cues but share the same backbone with the detector, e.g., a embedding head or a regression head.
 
-
 ### Add a new tracker
 
-
 #### 1. Define a tracker (e.g. MyTracker)
 
 Create a new file `mmtrack/models/mot/trackers/my_tracker.py`.
@@ -172,6 +170,7 @@ reid=dict(
 ```
 
 ### Add a new track head
+
 #### 1. Define a head (e.g. MyHead)
 
 Create a new file `mmtrack/models/track_heads/my_head.py`.
@@ -219,7 +218,6 @@ track_head=dict(
     arg2=xxx)
 ```
 
-
 ### Add a new loss
 
 Assume you want to add a new loss as `MyLoss`, for bounding box regression.

docs/tutorials/data_pipeline.md

@@ -5,14 +5,13 @@ There are two types of data pipelines in MMTracking:
 - Single image, which is consistent with MMDetection in most cases.
 - Pair-wise / multiple images.
 
-
 ### Data pipeline for a single image
 
 For a single image, you may refer to the [tutorial in MMDetection](https://mmdetection.readthedocs.io/en/latest/tutorials/data_pipeline.html).
 
 There are several differences in MMTracking:
-- We implement `VideoCollect` which is similar to `Collect` in MMDetection but is more comptabile with the video perception tasks.
-For example, the meta keys `frame_id` and `is_video_data` are collected by default.
+
+- We implement `VideoCollect` which is similar to `Collect` in MMDetection but is more comptabile with the video perception tasks. For example, the meta keys `frame_id` and `is_video_data` are collected by default.
 
 ### Data pipeline for multiple images
 
@@ -50,6 +49,7 @@ class CocoVideoDataset(CocoDataset):
             img_infos = self.ref_img_sampling(img_info, **self.ref_img_sampler)
         ...
 ```
+
 In this case, the loaded annotations is no longer a `dict` but `list[dict]` that contains the annotations for the key and reference images.
 The first item of the list indicates the annotations of the key image.
 
@@ -78,8 +78,8 @@ class LoadMultiImagesFromFile(LoadImageFromFile):
             outs.append(_results)
         return outs
 ```
-Sometimes you may need to add a parameter `share_params` to decide whether share the random seed of the transformation on these images.
 
+Sometimes you may need to add a parameter `share_params` to decide whether share the random seed of the transformation on these images.
 
 #### 3. Concat the reference images (if applicable)
 
@@ -90,8 +90,8 @@ The length of the list is 2 after the process.
 
 In the end, we implement `SeqDefaultFormatBundle` to convert the list to a dictionary as the input of the model forward.
 
-
 Here is an example of the data pipeline:
+
 ```python
 train_pipeline = [
     dict(type='LoadMultiImagesFromFile'),

docs/useful_tools.md

@@ -6,16 +6,19 @@ We provide lots of useful tools under the `tools/` directory.
 It is used as the same manner with `tools/test.py` but different in the configs.
 
 Here is an example that shows how to modify the configs:
+
 1. Define the desirable evaluation metrics to record.
 
     For example, you can define the search metrics as
+
     ```python
     search_metrics = ['MOTA', 'IDF1', 'FN', 'FP', 'IDs', 'MT', 'ML']
     ```
 
 2. Define the parameters and the values to search.
 
     Assume you have a tracker like
+
     ```python
     model = dict(
         tracker=dict(
@@ -27,6 +30,7 @@ Here is an example that shows how to modify the configs:
     ```
 
     If you want to search the parameters of the tracker, just change the value to a list as follow
+
     ```python
     model = dict(
         tracker=dict(