Add files via upload

INSTALL.md

@@ -0,0 +1,47 @@
+## Installation
+
+### Requirements:
+- Python3
+- PyTorch 1.1 with CUDA support
+- torchvision
+- pycocotools
+- yacs
+- matplotlib
+- GCC >= 4.9
+- (optional) OpenCV for the webcam demo
+
+
+### Step-by-step installation
+
+```bash
+# first, make sure that your conda is setup properly with the right environment
+# for that, check that `which conda`, `which pip` and `which python` points to the
+# right path. From a clean conda env, this is what you need to do
+
+conda create --name free_anchor python=3.7
+conda activate free_anchor
+
+# this installs the right pip and dependencies for the fresh python
+conda install ipython
+
+# maskrnn_benchmark and coco api dependencies
+pip install ninja yacs cython matplotlib tqdm
+
+# pytorch and torchvision
+# we give the instructions for CUDA 9.0
+conda install pytorch torchvision cudatoolkit=9.0 -c pytorch
+
+# install pycocotools
+pip install pycocotools
+
+# install PyTorch maskrcnn-benchmark.
+cd path_to_free_anchor
+# the following will install the lib with
+# symbolic links, so that you can modify
+# the files if you want and won't need to
+# re-build it
+python setup.py build develop
+
+# or if you are on macOS
+# MACOSX_DEPLOYMENT_TARGET=10.9 CC=clang CXX=clang++ python setup.py build develop
+```

README.md

@@ -0,0 +1,52 @@
+## FreeAnchor
+
+The code is based on the [maskrcnn-benchmark](https://github.com/facebookresearch/maskrcnn-benchmark).
+
+## Installation 
+Check [INSTALL.md](INSTALL.md) for installation instructions.
+
+## Perform Training on COCO dataset
+You will need to download the COCO dataset and configure your own paths to the datasets.
+
+For that, all you need to do is to modify `maskrcnn_benchmark/config/paths_catalog.py` to point to the location where your dataset is stored.
+
+### Config Files
+We provide four configuration files in the configs directory.
+
+| Backbone | Iter. | Setting | Config File |  
+| :-----: | :---: | :---: | :----------: |
+| ResNet-50-FPN    |   90k |   std.  | configs/free_anchor_R-50-FPN_1x.yaml      | 
+| ResNet-101-FPN   |   90k |   std.  | configs/free_anchor_R-101-FPN_1x.yaml     | 
+| ResNet-101-FPN   |  135k |   std.  | configs/free_anchor_R-101-FPN_1.5x.yaml   | 
+| ResNeXt-101-FPN  |  135k |   dev.  | configs/free_anchor_X-101-FPN_j1.5x.yaml  | 
+
+
+### 4 GPU (32GB memory) Training
+
+```bash
+cd path_to_free_anchor
+export NGPUS=4
+python -m torch.distributed.launch --nproc_per_node=$NGPUS tools/train_net.py --config-file "path/to/config/file.yaml"
+```
+
+### 8 GPU (>10GB memory) Training
+
+```bash
+cd path_to_free_anchor
+export NGPUS=8
+python -m torch.distributed.launch --nproc_per_node=$NGPUS tools/train_net.py --config-file "path/to/config/file.yaml"
+```
+
+### Test on MS-COCO test-dev
+
+```bash
+cd path_to_free_anchor
+python -m torch.distributed.launch --nproc_per_node=$NGPUS tools/test_net.py --config-file "path/to/config/file.yaml" MODEL.WEIGHT "path/to/.pth file" DATASETS.TEST "('coco_test-dev',)"
+```
+
+### Evaluate NMS Recall
+
+```bash
+cd path_to_free_anchor
+python  -m torch.distributed.launch --nproc_per_node=$NGPUS tools/eval_NR.py --config-file "path/to/config/file.yaml" MODEL.WEIGHT "path/to/.pth file"
+```

configs/free_anchor_R-101-FPN_1.5x.yaml

@@ -0,0 +1,47 @@
+MODEL:
+  META_ARCHITECTURE: "RetinaNet"
+  WEIGHT: "catalog://ImageNetPretrained/MSRA/R-101"
+  RPN_ONLY: True
+  BACKBONE:
+    CONV_BODY: "R-101-FPN"
+    OUT_CHANNELS: 256
+  RPN:
+    USE_FPN: True
+    ANCHOR_STRIDE: (4, 8, 16, 32, 64)
+    PRE_NMS_TOP_N_TRAIN: 2000
+    PRE_NMS_TOP_N_TEST: 1000
+    POST_NMS_TOP_N_TEST: 1000
+    FPN_POST_NMS_TOP_N_TEST: 1000
+    NMS_THRESH: 0.5
+DATASETS:
+  TRAIN: ("coco_2017_train",)
+  TEST: ("coco_2017_val",)
+INPUT:
+  MIN_SIZE_TRAIN: (800,)
+  MAX_SIZE_TRAIN: 1333
+  MIN_SIZE_TEST: 800
+  MAX_SIZE_TEST: 1333
+DATALOADER:
+  SIZE_DIVISIBILITY: 32
+SOLVER:
+  BASE_LR: 0.01
+  WEIGHT_DECAY: 0.0001
+  STEPS: (60000, 80000)
+  MAX_ITER: 90000
+  IMS_PER_BATCH: 16
+  WARMUP_ITERS: 500
+  CHECKPOINT_PERIOD: 2500
+RETINANET:
+  RETINANET_ON: True
+  NUM_CLASSES: 81
+  PRIOR_PROB: 0.02
+  SCALES_PER_OCTAVE: 3
+  STRADDLE_THRESH: -1
+FREEANCHOR:
+  FREEANCHOR_ON: True
+  PRE_ANCHOR_TOPK: 50
+  BBOX_THRESHOLD: 0.6
+  BBOX_REG_WEIGHT: 0.75
+  FOCAL_LOSS_ALPHA: 0.5
+  FOCAL_LOSS_GAMMA: 2.0
+OUTPUT_DIR: outputs/free_anchor_R-101-FPN_1x

configs/free_anchor_R-101-FPN_1x.yaml

@@ -0,0 +1,47 @@
+MODEL:
+  META_ARCHITECTURE: "RetinaNet"
+  WEIGHT: "catalog://ImageNetPretrained/MSRA/R-101"
+  RPN_ONLY: True
+  BACKBONE:
+    CONV_BODY: "R-101-FPN"
+    OUT_CHANNELS: 256
+  RPN:
+    USE_FPN: True
+    ANCHOR_STRIDE: (4, 8, 16, 32, 64)
+    PRE_NMS_TOP_N_TRAIN: 2000
+    PRE_NMS_TOP_N_TEST: 1000
+    POST_NMS_TOP_N_TEST: 1000
+    FPN_POST_NMS_TOP_N_TEST: 1000
+    NMS_THRESH: 0.5
+DATASETS:
+  TRAIN: ("coco_2017_train",)
+  TEST: ("coco_2017_val",)
+INPUT:
+  MIN_SIZE_TRAIN: (800,)
+  MAX_SIZE_TRAIN: 1333
+  MIN_SIZE_TEST: 800
+  MAX_SIZE_TEST: 1333
+DATALOADER:
+  SIZE_DIVISIBILITY: 32
+SOLVER:
+  BASE_LR: 0.01
+  WEIGHT_DECAY: 0.0001
+  STEPS: (90000, 120000)
+  MAX_ITER: 135000
+  IMS_PER_BATCH: 16
+  WARMUP_ITERS: 500
+  CHECKPOINT_PERIOD: 2500
+RETINANET:
+  RETINANET_ON: True
+  NUM_CLASSES: 81
+  PRIOR_PROB: 0.02
+  SCALES_PER_OCTAVE: 3
+  STRADDLE_THRESH: -1
+FREEANCHOR:
+  FREEANCHOR_ON: True
+  PRE_ANCHOR_TOPK: 50
+  BBOX_THRESHOLD: 0.6
+  BBOX_REG_WEIGHT: 0.75
+  FOCAL_LOSS_ALPHA: 0.5
+  FOCAL_LOSS_GAMMA: 2.0
+OUTPUT_DIR: outputs/free_anchor_R-101-FPN_1.5x

configs/free_anchor_R-50-FPN_1x.yaml

@@ -0,0 +1,47 @@
+MODEL:
+  META_ARCHITECTURE: "RetinaNet"
+  WEIGHT: "catalog://ImageNetPretrained/MSRA/R-50"
+  RPN_ONLY: True
+  BACKBONE:
+    CONV_BODY: "R-50-FPN"
+    OUT_CHANNELS: 256
+  RPN:
+    USE_FPN: True
+    ANCHOR_STRIDE: (4, 8, 16, 32, 64)
+    PRE_NMS_TOP_N_TRAIN: 2000
+    PRE_NMS_TOP_N_TEST: 1000
+    POST_NMS_TOP_N_TEST: 1000
+    FPN_POST_NMS_TOP_N_TEST: 1000
+    NMS_THRESH: 0.5
+DATASETS:
+  TRAIN: ("coco_2017_train",)
+  TEST: ("coco_2017_val",)
+INPUT:
+  MIN_SIZE_TRAIN: (800,)
+  MAX_SIZE_TRAIN: 1333
+  MIN_SIZE_TEST: 800
+  MAX_SIZE_TEST: 1333
+DATALOADER:
+  SIZE_DIVISIBILITY: 32
+SOLVER:
+  BASE_LR: 0.01
+  WEIGHT_DECAY: 0.0001
+  STEPS: (60000, 80000)
+  MAX_ITER: 90000
+  IMS_PER_BATCH: 16
+  WARMUP_ITERS: 500
+  CHECKPOINT_PERIOD: 2500
+RETINANET:
+  RETINANET_ON: True
+  NUM_CLASSES: 81
+  PRIOR_PROB: 0.02
+  SCALES_PER_OCTAVE: 3
+  STRADDLE_THRESH: -1
+FREEANCHOR:
+  FREEANCHOR_ON: True
+  PRE_ANCHOR_TOPK: 50
+  BBOX_THRESHOLD: 0.6
+  BBOX_REG_WEIGHT: 0.75
+  FOCAL_LOSS_ALPHA: 0.5
+  FOCAL_LOSS_GAMMA: 2.0
+OUTPUT_DIR: outputs/free_anchor_R-50-FPN_1x

configs/free_anchor_X-101-FPN_j1.5x.yaml

@@ -0,0 +1,51 @@
+MODEL:
+  META_ARCHITECTURE: "RetinaNet"
+  WEIGHT: "catalog://ImageNetPretrained/FAIR/20171220/X-101-32x8d"
+  RPN_ONLY: True
+  BACKBONE:
+    CONV_BODY: "R-101-FPN"
+    OUT_CHANNELS: 256
+  RESNETS:
+    STRIDE_IN_1X1: False
+    NUM_GROUPS: 32
+    WIDTH_PER_GROUP: 8
+  RPN:
+    USE_FPN: True
+    ANCHOR_STRIDE: (4, 8, 16, 32, 64)
+    PRE_NMS_TOP_N_TRAIN: 2000
+    PRE_NMS_TOP_N_TEST: 1000
+    POST_NMS_TOP_N_TEST: 1000
+    FPN_POST_NMS_TOP_N_TEST: 1000
+    NMS_THRESH: 0.5
+DATASETS:
+  TRAIN: ("coco_2017_train",)
+  TEST: ("coco_2017_val",)
+INPUT:
+  MIN_SIZE_TRAIN: (640, 672, 704, 736, 768, 800,)
+  MAX_SIZE_TRAIN: 1333
+  MIN_SIZE_TEST: 800
+  MAX_SIZE_TEST: 1333
+DATALOADER:
+  SIZE_DIVISIBILITY: 32
+SOLVER:
+  BASE_LR: 0.01
+  WEIGHT_DECAY: 0.0001
+  STEPS: (90000, 120000)
+  MAX_ITER: 135000
+  IMS_PER_BATCH: 16
+  WARMUP_ITERS: 500
+  CHECKPOINT_PERIOD: 2500
+RETINANET:
+  RETINANET_ON: True
+  NUM_CLASSES: 81
+  PRIOR_PROB: 0.02
+  SCALES_PER_OCTAVE: 3
+  STRADDLE_THRESH: -1
+FREEANCHOR:
+  FREEANCHOR_ON: True
+  PRE_ANCHOR_TOPK: 50
+  BBOX_THRESHOLD: 0.6
+  BBOX_REG_WEIGHT: 0.75
+  FOCAL_LOSS_ALPHA: 0.5
+  FOCAL_LOSS_GAMMA: 2.0
+OUTPUT_DIR: outputs/free_anchor_X-101-FPN_j1.5x

demo/README.md

@@ -0,0 +1,16 @@
+## Webcam and Jupyter notebook demo
+
+This folder contains a simple webcam demo that illustrates how you can use `maskrcnn_benchmark` for inference.
+
+You can start it by running it from this folder, using one of the following commands:
+```bash
+# by default, it runs on the GPU
+# for best results, use min-image-size 800
+python webcam.py --min-image-size 800
+# can also run it on the CPU
+python webcam.py --min-image-size 300 MODEL.DEVICE cpu
+# or change the model that you want to use
+python webcam.py --config-file ../configs/caffe2/e2e_mask_rcnn_R_101_FPN_1x_caffe2.py --min-image-size 300 MODEL.DEVICE cpu
+# in order to see the probability heatmaps, pass --show-mask-heatmaps
+python webcam.py --min-image-size 300 --show-mask-heatmaps MODEL.DEVICE cpu
+```