Update readme with AutoGluon (#1766)

* introduce ag

* fix opencv

* fix

* fix

* test

* test

* test

* test

* fix

* test

* fix six

* address comments

Co-authored-by: Weisu Yin <[email protected]>

Author: bryanyzhu

.github/workflows/unittest.yml

@@ -39,17 +39,9 @@ jobs:
         shell: bash -l {0}
         run: |
           conda env create -n gluon_cv_py3_test -f tests/py3_mxnet_ci.yml
-          conda env update -n gluon_cv_py3_test -f tests/py3_mxnet_ci.yml --prune
           conda activate gluon_cv_py3_test
-          conda list
-          export CUDA_VISIBLE_DEVICES=0
-          export KMP_DUPLICATE_LIB_OK=TRUE
-          make clean
-          pip install --upgrade --force-reinstall --no-deps .
-          env
-          export LD_LIBRARY_PATH=/usr/local/cuda-10.0/lib64
-          export MPLBACKEND=Agg
-          export MXNET_CUDNN_AUTOTUNE_DEFAULT=0
+          pip install --upgrade pip setuptools wheel
+          pip install --upgrade -e .
           export TINY_COCO=~/.mxnet/datasets/tiny_coco
           export TINY_MOTORBIKE=~/.mxnet/datasets/tiny_motorbike
           mkdir -p $TINY_COCO/annotations

README.md

@@ -26,6 +26,9 @@ models. This toolkit offers four main features:
 4. Carefully designed APIs that greatly reduce the implementation complexity
 5. Community supports
 
+Please also checkout [AutoGluon](https://github.com/autogluon/autogluon) if you have [image classification](https://auto.gluon.ai/stable/tutorials/multimodal/image_prediction/index.html) or [object detection](https://auto.gluon.ai/stable/tutorials/multimodal/object_detection/index.html) needs. We have built the [MultimodalPredictor](https://auto.gluon.ai/stable/tutorials/multimodal/index.html) with an improved model zoo, including [TIMM](https://github.com/rwightman/pytorch-image-models), [Huggingface](https://huggingface.co/), [MMDetection](https://github.com/open-mmlab/mmdetection) and more. With just a few lines of code, you can train and deploy high accuracy computer vision models for your application.
+
+
 # Demo
 
 <div align="center">

docs/tutorials/index.rst

@@ -3,6 +3,10 @@ Apache MXNet Tutorials
 
 Interested in getting started in a new CV area? Here are some tutorials to help get started.
 
+Note: For image classification or object detection tasks, feel free to explore tutorials in
+`AutoGluon MultiModalPredictor <https://auto.gluon.ai/stable/tutorials/multimodal/index.html>`_ with better support in PyTorch.
+
+
 Image Classification
 --------------------
 

gluoncv/torch/data/video_cls/multigrid_helper.py

@@ -3,7 +3,7 @@
 import numpy as np
 
 from torch.utils.data import Sampler
-from torch._six import int_classes as _int_classes
+_int_classes = int
 
 
 __all__ = ['multiGridHelper', 'MultiGridBatchSampler']

tests/py3_mxnet_ci.yml

@@ -3,26 +3,10 @@ channels:
   - conda-forge
   - defaults
 dependencies:
-  - python=3.6
-  - perl
-  - sphinx=1.7.2
+  - python=3.7
   - nose
   - coverage=4.5.4
-  - scipy
-  - cython
-  - pip=20.2.4
-  - requests==2.27.1
-  - matplotlib
-  - tqdm
-  - pillow
   - pip:
     - mxnet
     - coverage-badge
-    - awscli
     - nose-timer
-    - opencv-python
-    - git+https://github.com/zhanghang1989/detail-api.git#subdirectory=PythonAPI
-    - portalocker
-    - autocfg>=0.0.6
-    - boto3
-    - autogluon.core==0.2.0