Skip to content

S
seminar-breakout
Commit 768b835e authored by Yuxin Wu's avatar Yuxin Wu
Browse files
Options

udpate docs

parent 5feb3529
Hide whitespace changes
Inline Side-by-side
Showing with 14 additions and 65 deletions
examples/FasterRCNN/README.md
View file @ 768b835e
# Faster R-CNN / Mask R-CNN on COCO
This example provides a minimal (2k lines) and faithful implementation of the following papers:
This example provides a minimal (2k lines) and faithful implementation of the
following object detection / instance segmentation papers:
+ [Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks](https://arxiv.org/abs/1506.01497)
+ [Feature Pyramid Networks for Object Detection](https://arxiv.org/abs/1612.03144)
......@@ -13,11 +14,11 @@ with the support of:
+ Training from scratch (from [Rethinking ImageNet Pre-training](https://arxiv.org/abs/1811.08883))
## Dependencies
+ Python 3.3+; OpenCV.
+ TensorFlow ≥ 1.6 (1.4 or 1.5 can run but may crash due to a TF bug);
+ Python 3.3+; OpenCV
+ TensorFlow ≥ 1.6
+ pycocotools: `pip install 'git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI'`
+ Pre-trained [ImageNet ResNet model](http://models.tensorpack.com/FasterRCNN/)
from tensorpack model zoo.
from tensorpack model zoo
+ [COCO data](http://cocodataset.org/#download). It needs to have the following directory structure:
```
COCO/DIR/
......@@ -78,15 +79,17 @@ These models are trained on trainval35k and evaluated on minival2014 using mAP@I
All models are fine-tuned from ImageNet pre-trained R50/R101 models in
[tensorpack model zoo](http://models.tensorpack.com/FasterRCNN/), unless otherwise noted.
All models are trained with 8 NVIDIA V100s, unless otherwise noted.
Performance in [Detectron](https://github.com/facebookresearch/Detectron/) can be roughly reproduced.
FPN models are sometimes slightly worse, which is mainly due to batch size.
| Backbone | mAP<br/>(box;mask) | Detectron mAP <sup>[1](#ft1)</sup><br/> (box;mask) | Time (on 8 V100s) | Configurations <br/> (click to expand) |
| - | - | - | - | - |
| R50-C4 | 33.1 | | 18h | <details><summary>super quick</summary>`MODE_MASK=False FRCNN.BATCH_PER_IM=64`<br/>`PREPROC.SHORT_EDGE_SIZE=600 PREPROC.MAX_SIZE=1024`<br/>`TRAIN.LR_SCHEDULE=[150000,230000,280000]` </details> |
| R50-C4 | 36.6 | 36.5 | 44h | <details><summary>standard</summary>`MODE_MASK=False` </details> |
| R50-FPN | 37.4 | 37.9 | 29h | <details><summary>standard</summary>`MODE_MASK=False MODE_FPN=True` </details> |
| R50-C4 | 38.2;33.3 [:arrow_down:](http://models.tensorpack.com/FasterRCNN/COCO-R50C4-MaskRCNN-Standard.npz) | 37.8;32.8 | 49h | <details><summary>standard</summary>this is the default </details> |
| R50-FPN | 38.5;35.2 [:arrow_down:](http://models.tensorpack.com/FasterRCNN/COCO-R50FPN-MaskRCNN-Standard.npz) | 38.6;34.5 | 30h | <details><summary>standard</summary>`MODE_FPN=True` </details> |
| R50-FPN | 37.4 | 37.9 | 27h | <details><summary>standard</summary>`MODE_MASK=False MODE_FPN=True` </details> |
| R50-C4 | 38.2;33.3 [:arrow_down:](http://models.tensorpack.com/FasterRCNN/COCO-R50C4-MaskRCNN-Standard.npz) | 37.8;32.8 | 48h | <details><summary>standard</summary>this is the default </details> |
| R50-FPN | 38.4;35.1 [:arrow_down:](http://models.tensorpack.com/FasterRCNN/COCO-R50FPN-MaskRCNN-Standard.npz) | 38.6;34.5 | 28h | <details><summary>standard</summary>`MODE_FPN=True` </details> |
| R50-FPN | 42.0;36.3 | | 41h | <details><summary>+Cascade</summary>`MODE_FPN=True FPN.CASCADE=True` </details> |
| R50-FPN | 39.5;35.2 | 39.5;34.4<sup>[2](#ft2)</sup> | 33h | <details><summary>+ConvGNHead</summary>`MODE_FPN=True`<br/>`FPN.FRCNN_HEAD_FUNC=fastrcnn_4conv1fc_gn_head` </details> |
| R50-FPN | 40.0;36.2 [:arrow_down:](http://models.tensorpack.com/FasterRCNN/COCO-R50FPN-MaskRCNN-StandardGN.npz) | 40.3;35.7 | 40h | <details><summary>+GN</summary>`MODE_FPN=True`<br/>`FPN.NORM=GN BACKBONE.NORM=GN`<br/>`FPN.FRCNN_HEAD_FUNC=fastrcnn_4conv1fc_gn_head`<br/>`FPN.MRCNN_HEAD_FUNC=maskrcnn_up4conv_gn_head` |
......
tensorpack/graph_builder/model_desc.py
View file @ 768b835e
......@@ -37,8 +37,7 @@ class InputDesc(
self._cached_placeholder = {}
return self
# TODO this method seems unused outside this class
def build_placeholder(self):
def _build_placeholder(self):
"""
Build a tf.placeholder from the metadata.
......@@ -63,7 +62,7 @@ class InputDesc(
if g in self._cached_placeholder:
return self._cached_placeholder[g]
else:
return self.build_placeholder()
return self._build_placeholder()
def _register_cached_placeholder(self, placeholder):
graph = placeholder.graph
......@@ -241,12 +240,13 @@ class ModelDesc(ModelDescBase):
def _build_graph_get_cost(self, *inputs):
"""
Equivalent to `build_graph`.
Used internally by trainers to get the final cost for optimization in a backward-compatible way.
"""
ret = self.build_graph(*inputs)
if not get_current_tower_context().is_training:
return None # this is the tower function, could be called for inference
if isinstance(ret, tf.Tensor): # the preferred way
if ret is not None:
return ret
else: # the old way, for compatibility
return self.get_cost()
tensorpack/utils/argtools.py
View file @ 768b835e
......@@ -75,8 +75,6 @@ def memoized_ignoreargs(func):
A decorator. It performs memoization ignoring the arguments used to call
the function.
"""
hash(func) # make sure it is hashable. TODO is it necessary?
def wrapper(*args, **kwargs):
if func not in _MEMOIZED_NOARGS:
res = func(*args, **kwargs)
......
tensorpack/utils/globvars.py deleted 100644 → 0
View file @ 5feb3529
# -*- coding: utf-8 -*-
# File: globvars.py
import six
import argparse
from . import logger
__all__ = ['globalns', 'GlobalNS']
if six.PY2:
class NS:
pass
else:
import types
NS = types.SimpleNamespace
# TODO make it singleton
class GlobalNS(NS):
"""
The class of the globalns instance.
"""
def use_argument(self, args):
"""
Add the content of :class:`argparse.Namespace` to this ns.
Args:
args (argparse.Namespace): arguments
"""
assert isinstance(args, argparse.Namespace), type(args)
for k, v in six.iteritems(vars(args)):
if hasattr(self, k):
logger.warn("Attribute {} in globalns will be overwritten!")
setattr(self, k, v)
globalns = GlobalNS()
"""
A namespace to store global variables.
Example:
.. code-block:: none
import tensorpack.utils.globalns as G
G.depth = 18
G.batch_size = 1
G.use_argument(parser.parse_args())
"""
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment