Commits · 70dee77760b88d2ff82dc27647a430fc6d3f7c7b · OpenDAS / vision

22 Jan, 2020 4 commits
- Update README to reflect strict dependency on torch==1.4.0 (#1767) · 70dee777
  Danny Vilela authored Jan 22, 2020
```
* Update README to reflect strict dependency on torch==1.4.0

* 1.4 or newer.
```
  70dee777
- fixed function description (#1768) · 0c92f9c7
  Ebey Abraham authored Jan 23, 2020
  
  0c92f9c7
- Fix Windows build by renaming Python init functions (#1779) · e2a8b418
  peterjc123 authored Jan 23, 2020
  
  e2a8b418
- Fix fill in rotate (#1760) · 8f4f8d89
  Philip Meier authored Jan 22, 2020
```
* initial fix

* outsourced num bands lookup

* fix doc

* added pillow version requirement

* simplify number of bands extraction

* remove unrelated change

* remove indirect dependency on pillow>=5.2.0

* extend docstring to transform

* bug fix

* added test
```
  8f4f8d89
21 Jan, 2020 1 commit
- typing only needed for python 3.5 and previous (#1778) · d9a3018f
  Jeremy Reizenstein authored Jan 21, 2020
  
  d9a3018f
20 Jan, 2020 1 commit
- Make C++ build with CUDA support optional (#1772) · cc404108
  keith authored Jan 21, 2020
  
  cc404108
17 Jan, 2020 4 commits

Remove unintentional -O0 option in setup.py (#1770) · 9a22a8a7

Richard Zou authored Jan 17, 2020

Previously, when doing a CUDA build, we would pass -O0 to build cpu
bits. This PR removes that `-O0` (so we build in `-O3` instead.

Test Plan:
- wait for CI

9a22a8a7

Fix C++ API build work with libtorch (#1764) · f9229160
keith authored Jan 17, 2020

f9229160

ci: Remove Python 2.7 (#1761) · 885db4b1

Eli Uriegas authored Jan 17, 2020



Python 2.7 was EOL on January 1, 2020

The last torchvision release to support Python 2.7 was 0.5.0
Signed-off-by: Eli Uriegas <eliuriegas@fb.com>

885db4b1

Changes in docstring example for RCNN based models. (#1763) · bbaa6f3b
Marco Martinelli authored Jan 17, 2020
```
* Type of input featmap_names fixed in example.

* Added missing imports.
```
bbaa6f3b

16 Jan, 2020 2 commits
- UIpdate Doc with ONNX support (#1752) · c5e972a6
  Lara Haidar authored Jan 16, 2020
```
* update doc

* update doc
```
  c5e972a6
- CUDA_SUFFIX -> PYTORCH_VERSION_SUFFIX · add2ecc7
  Soumith Chintala authored Jan 15, 2020
  
  add2ecc7
14 Jan, 2020 2 commits
- Fixes EMNIST classes attribute is wrong #1716 (#1736) · 32df87a8
  Gokkulnath TS authored Jan 14, 2020
```
* Fixes #1716

Fixes EMNIST classes attribute is wrong #1716

* Fixed the Classes for Letters Split

* Update mnist.py

* Move classes attribute inside init definition

* Fix Linting errors
```
  32df87a8
- master version bump 0.5 -> 0.6 · a199bf62
  Soumith Chintala authored Jan 13, 2020
  
  a199bf62
13 Jan, 2020 2 commits
- Fix for AnchorGenerator when device switch happen (#1745) · 66f29220
  Francisco Massa authored Jan 13, 2020
```
* Fix AnchorGenerator if moving from one device to another

* Fixes for the test
```
  66f29220
- Remove constants from DenseBlock (#1727) · 7bfbc81a
  Francisco Massa authored Jan 13, 2020
  
  7bfbc81a
10 Jan, 2020 2 commits
- fix for loading models with num_batches_tracked in frozen bn (#1728) · 61763fa9
  Sergey Zagoruyko authored Jan 10, 2020
  
  61763fa9
- Testing CI (#1735) · be6dd472
  Francisco Massa authored Jan 10, 2020
```
* Testing CI

* Disable tests for Pillow 7
```
  be6dd472
03 Jan, 2020 1 commit
- Update quantized shufflenet weights (#1715) · 07cbb46a
  Francisco Massa authored Jan 03, 2020
```
Previous weights are not compatible with current PyTorch
```
  07cbb46a
02 Jan, 2020 3 commits

Fix lint following #1695 (#1713) · 300019dd
Francisco Massa authored Jan 02, 2020
```
* Fix lint following #1695

* V2

* V3
```
300019dd

Speed up nms_cuda (#1704) · 06cbdb5b

Yuxin Wu authored Jan 02, 2020

1. Let the IOU function compare with threshold. This avoid a division. Similar strategy is also used in https://github.com/tensorflow/tensorflow/blob/master/tensorflow/core/kernels/non_max_suppression_op.cu.cc
2. Only compute the upper triangle of the mask.

This speeds up the kernel about 20% (tested on GTX 1080Ti, with 20 input cases dumped from a Mask R-CNN inference job).

06cbdb5b

Added Training Sample code for fasterrcnn_resnet50_fpn (#1695) · 2a174229
Prajjwal Bhargava authored Jan 02, 2020

2a174229

19 Dec, 2019 6 commits
- Fix lint (#1693) · d2c763e1
  Francisco Massa authored Dec 19, 2019
  
  d2c763e1
- Fix lint (#1692) · 013d1322
  Francisco Massa authored Dec 19, 2019
  
  013d1322
- fix a little bug about resume (#1628) · f4a82243
  MultiK authored Dec 20, 2019
```
* fix a little bug about resume

When resuming, we need to start from the last epoch not 0.

* the second way for resuming

the second way for resuming
```
  f4a82243
- Scriptability checks for Tensor Transforms (#1690) · 10f34160
  Surgan Jandial authored Dec 20, 2019
```
* scriptability checks

* tests upds

* linter upds

* linter upds

* upds

* tuple list changes

* linter updates
```
  10f34160
- Bugfix on GroupedBatchSampler for corner case where there are not enough... · 900c88c7
  Francisco Massa authored Dec 19, 2019
```
Bugfix on GroupedBatchSampler for corner case where there are not enough examples in a category to form a batch (#1677)
```
  900c88c7
- Enable KeypointRCNN test (#1673) · 1d229b77
  Lara Haidar authored Dec 19, 2019
  
  1d229b77
17 Dec, 2019 1 commit
- Make R-CNN models less verbose in script mode (#1671) · d32bce08
  Francisco Massa authored Dec 17, 2019
```
* Make R-CNN models less verbose in script mode

* Fix typo in warning message
```
  d32bce08
16 Dec, 2019 4 commits
- Bugfix in BalancedPositiveNegativeSampler introduced during torchscript support (#1670) · c7c2085e
  Francisco Massa authored Dec 16, 2019
  
  c7c2085e
- fixed test for windows by closing the created temporary files (#1662) · bce17fdd
  Oana Florescu authored Dec 16, 2019
  
  bce17fdd
- VideoClips windows fixes (#1661) · 757ecfb0
  Oana Florescu authored Dec 16, 2019
```
* remove windows skips from video_utils tests, now that they pass

* replace lambda in videoclips in order to be pickled on windows and update tests
```
  757ecfb0
- Improve error message and avoid segfault in DeformConv2d (#1660) · 17ea1482
  Francisco Massa authored Dec 16, 2019
  
  17ea1482
12 Dec, 2019 1 commit
- Test for checking non mutating behaviour of tensor transforms (#1656) · 2d7c0667
  Surgan Jandial authored Dec 12, 2019
```
* out_place checks

* lint ups
```
  2d7c0667
11 Dec, 2019 1 commit
- Fix rpn memory leak and dataType errors. (#1657) · 5c03d593
  TengQi Ye authored Dec 11, 2019
  
  5c03d593
10 Dec, 2019 2 commits
- Move VideoClips dummy dataset to top level for pickling (#1649) · 333af7aa
  James Thewlis authored Dec 10, 2019
  
  333af7aa
- add .tgz support to extract_archive (#1650) · 4ab94cb7
  Surgan Jandial authored Dec 10, 2019
```
* tgz updates

* tgz updates

* tgz updates
```
  4ab94cb7
06 Dec, 2019 1 commit
- Fix header includes for cpu (#1644) · 598b61d9
  gslotman authored Dec 06, 2019
  
  598b61d9
05 Dec, 2019 2 commits
- PyTorch has switched to C++14. (#1635) · 3bd7a405
  xkszltl authored Dec 05, 2019
  
  3bd7a405
- [BC-breaking] Update KeypointRCNN weights (#1609) · 6bf920b9
  Francisco Massa authored Dec 05, 2019
```
* Update KeypointRCNN weights with correct file

* Fix model

* Fix
```
  6bf920b9