update the code to python3 and numpy 1.12.1

LICENSE

@@ -32,7 +32,7 @@ notice and license of such Third Party Code are set out below. This
 Third Party Code is licensed to you under their original license terms 
 set forth below.
 
-1. Fast R-CNN (https://github.com/rbgirshick/fast-rcnn)
+1.Fast R-CNN (https://github.com/rbgirshick/fast-rcnn)
 
 Copyright (c) Microsoft Corporation
 
@@ -58,32 +58,7 @@ OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
 ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
 OTHER DEALINGS IN THE SOFTWARE.
 
-
-2. Faster R-CNN (https://github.com/rbgirshick/py-faster-rcnn)
-
-The MIT License (MIT)
-
-Copyright (c) 2015 Microsoft Corporation
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in
-all copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
-THE SOFTWARE.
-
-3.	Caffe, (https://github.com/BVLC/caffe/)
+2.Caffe, (https://github.com/BVLC/caffe/)
 
 COPYRIGHT
 

README.md

@@ -37,8 +37,12 @@ This implementation is built on a *fork* of the OHEM code ([here](https://github
 
 
 ### Installation
-
-Please follow the exact installation and download the VOC data as the Faster R-CNN Python code ([here](https://github.com/rbgirshick/py-faster-rcnn)).
+- Step 1:
+  Download the selective search data by running script 'data/script/fetch_selective_search_data.sh'.
+- Step 2:
+  Download the caffemodel by running script 'data/script/fetch_imagenet_models.sh'.
+- Step 3:
+  Create the link to VOCdevkit.You can find the details from ['data/README.md'](https://github.com/NB-Dragon/A-Fast-RCNN/blob/master/data/README.md).
 
 ### Usage
 
@@ -50,21 +54,9 @@ To run the code, one can simply do,
 It includes 3-stage of training:
 
 ```Shell
-./experiments/scripts/fast_rcnn_std.sh  [GPU_ID]  VGG16 pascal_voc
-```
-which is used for training a standard Fast-RCNN for 10K iterations, you can download my [model](https://www.dropbox.com/s/ccs7lw3gydfzgvv/fast_rcnn_std_iter_10000.caffemodel?dl=0) and [logs](https://www.dropbox.com/s/hwbag60l1gmtxbb/fast_rcnn_std.txt.2017-04-08_16-53-59?dl=0) for this step.
-
-```Shell
-./experiments/scripts/fast_rcnn_adv_pretrain.sh  [GPU_ID]  VGG16 pascal_voc
-```
-which is a pre-training stage for the adversarial network, you can download my [model](https://www.dropbox.com/s/hvqpxn3bigarhdn/fast_rcnn_adv_pretrain_iter_25000.caffemodel?dl=0) and [logs](https://www.dropbox.com/s/i79j5hd0ee4ybke/fast_rcnn_adv_pretrain.txt.2017-04-08_19-39-49?dl=0) for this step.
-
-```Shell
+./experiments/scripts/fast_rcnn_std.sh [GPU_ID] VGG16 pascal_voc
+./experiments/scripts/fast_rcnn_adv_pretrain.sh [GPU_ID] VGG16 pascal_voc
 ./copy_model.h
+./experiments/scripts/fast_rcnn_adv.sh [GPU_ID] VGG16 pascal_voc
 ```
-which is used to copy the weights of the above two models to initialize the joint model.
-
-```Shell
-./experiments/scripts/fast_rcnn_adv.sh  [GPU_ID]  VGG16 pascal_voc
-```
-which is joint training of the detector and the adversarial network, you can download my [model](https://www.dropbox.com/s/5wvxh8g5n3ewvp4/fast_rcnn_adv_iter_40000.caffemodel?dl=0) and [logs](https://www.dropbox.com/s/awrdrwyfthdgba5/fast_rcnn_adv.txt.2017-04-09_22-09-57?dl=0) for this step.
+All output training files and logs can be downloaded from my [Baidu cloud disk](https://pan.baidu.com/s/1nvac2Jv).

data/README.md

@@ -1,25 +1,33 @@
 This directory holds (*after you download them*):
-- Fast R-CNN models trained with OHEM on VOC 2007 trainval
+- Pre-computed object proposals
 - Caffe models pre-trained on ImageNet
+- Fast R-CNN models
 - Symlinks to datasets
 
-To download Fast R-CNN models (VGG_CNN_M_1024, VGG16) trained with OHEM on VOC 2007 trainval, run:
+To download precomputed Selective Search proposals for PASCAL VOC 2007 and 2012, run:
 
 ```
-./data/scripts/fetch_fast_rcnn_ohem_models.sh
+./data/scripts/fetch_selective_search_data.sh
 ```
 
-This script will populate `data/fast_rcnn_ohem_models` with VGG16 and VGG_CNN_M_1024 models (Fast R-CNN detectors trained with OHEM).
+This script will populate `data/selective_search_data`.
 
-
-To download Caffe models (ZF, VGG16) pre-trained on ImageNet, run:
+To download Caffe models (CaffeNet, VGG_CNN_M_1024, VGG16) pre-trained on ImageNet, run:
 
 ```
 ./data/scripts/fetch_imagenet_models.sh
 ```
 
 This script will populate `data/imagenet_models`.
 
+To download Fast R-CNN models trained on VOC 2007, run:
+
+```
+./data/scripts/fetch_fast_rcnn_models.sh
+```
+
+This script will populate `data/fast_rcnn_models`.
+
 In order to train and test with PASCAL VOC, you will need to establish symlinks.
 From the `data` directory (`cd data`):
 
@@ -31,29 +39,10 @@ ln -s /your/path/to/VOC2007/VOCdevkit VOCdevkit2007
 ln -s /your/path/to/VOC2012/VOCdevkit VOCdevkit2012
 ```
 
-Install the MS COCO dataset at /path/to/coco
-
-```
-ln -s /path/to/coco coco
-```
-
-For COCO with Fast R-CNN, place object proposals under `coco_proposals` (inside
-the `data` directory). You can obtain proposals on COCO from Jan Hosang at
-https://www.mpi-inf.mpg.de/departments/computer-vision-and-multimodal-computing/research/object-recognition-and-scene-understanding/how-good-are-detection-proposals-really/.
-For COCO, using MCG is recommended over selective search. MCG boxes can be downloaded
-from http://www.eecs.berkeley.edu/Research/Projects/CS/vision/grouping/mcg/.
-Use the tool `lib/datasets/tools/mcg_munge.py` to convert the downloaded MCG data
-into the same file layout as those from Jan Hosang.
-
-Since you'll likely be experimenting with multiple installs of Fast/er R-CNN in
+Since you'll likely be experimenting with multiple installs of Fast R-CNN in
 parallel, you'll probably want to keep all of this data in a shared place and
 use symlinks. On my system I create the following symlinks inside `data`:
 
-Annotations for the 5k image 'minival' subset of COCO val2014 that I like to use
-can be found at http://www.cs.berkeley.edu/~rbg/faster-rcnn-data/instances_minival2014.json.zip.
-Annotations for COCO val2014 (set) minus minival (~35k images) can be found at
-http://www.cs.berkeley.edu/~rbg/faster-rcnn-data/instances_valminusminival2014.json.zip.
-
 ```
 # data/cache holds various outputs created by the datasets package
 ln -s /data/fast_rcnn_shared/cache
@@ -62,7 +51,6 @@ ln -s /data/fast_rcnn_shared/cache
 ln -s /data/fast_rcnn_shared/imagenet_models
 
 # move the selective search data to a shared location and symlink to them
-# (only applicable to Fast R-CNN training)
 ln -s /data/fast_rcnn_shared/selective_search_data
 
 ln -s /data/VOC2007/VOCdevkit VOCdevkit2007

data/scripts/fetch_fast_rcnn_ohem_models.sh → data/scripts/fetch_fast_rcnn_models.sh

@@ -3,9 +3,9 @@
 DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )/../" && pwd )"
 cd $DIR
 
-FILE=fast_rcnn_ohem_models.tgz
-URL=http://graphics.cs.cmu.edu/projects/ohem/data/$FILE
-CHECKSUM=cbfd5b7ed5ec4d5cb838701cbf1f3ccb
+FILE=fast_rcnn_models.tgz
+URL=https://dl.dropboxusercontent.com/s/e3ugqq3lca4z8q6/fast_rcnn_models.tgz
+CHECKSUM=5f7dde9f5376e18c8e065338cc5df3f7
 
 if [ -f $FILE ]; then
   echo "File already exists. Checking md5..."
@@ -23,7 +23,7 @@ if [ -f $FILE ]; then
   fi
 fi
 
-echo "Downloading Fast R-CNN OHEM models (VGG16 and VGG_CNN_M_1024)(1.5G)..."
+echo "Downloading Fast R-CNN demo models (0.96G)..."
 
 wget $URL -O $FILE
 

data/scripts/fetch_imagenet_models.sh

@@ -4,8 +4,8 @@ DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )/../" && pwd )"
 cd $DIR
 
 FILE=imagenet_models.tgz
-URL=http://www.cs.berkeley.edu/~rbg/faster-rcnn-data/$FILE
-CHECKSUM=ed34ca912d6782edfb673a8c3a0bda6d
+URL=https://dl.dropboxusercontent.com/s/riazjuizq0w7dqm/imagenet_models.tgz
+CHECKSUM=8b1d4b9da0593fc70ef403284f810adc
 
 if [ -f $FILE ]; then
   echo "File already exists. Checking md5..."

data/scripts/fetch_selective_search_data.sh

@@ -3,9 +3,9 @@
 DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )/../" && pwd )"
 cd $DIR
 
-FILE=selective_search_data.tgz
-URL=http://www.cs.berkeley.edu/~rbg/fast-rcnn-data/$FILE
-CHECKSUM=7078c1db87a7851b31966b96774cd9b9
+FILE=r-cnn-release1-selective-search.tgz
+URL=https://dl.dropboxusercontent.com/s/uf2i1y2oee7c6n1/r-cnn-release1-selective-search.tgz
+CHECKSUM=6cf6df219c1e514f64482f11d00bd0b4
 
 if [ -f $FILE ]; then
   echo "File already exists. Checking md5..."
@@ -23,7 +23,7 @@ if [ -f $FILE ]; then
   fi
 fi
 
-echo "Downloading precomputed selective search boxes (0.5G)..."
+echo "Downloading precomputed selective search boxes (1.8G)..."
 
 wget $URL -O $FILE
 

lib/datasets/VOCdevkit-matlab-wrapper/voc_eval.m

@@ -2,6 +2,7 @@
 
 VOCopts = get_voc_opts(path);
 VOCopts.testset = test_set;
+VOCopts.detrespath=[VOCopts.resdir 'Main/%s_det_' VOCopts.testset '_%s.txt'];
 
 for i = 1:length(VOCopts.classes)
   cls = VOCopts.classes{i};

lib/datasets/coco.py

@@ -13,7 +13,7 @@
 import numpy as np
 import scipy.sparse
 import scipy.io as sio
-import cPickle
+import pickle
 import json
 import uuid
 # COCO API
@@ -33,7 +33,7 @@ def _filter_crowd_proposals(roidb, crowd_thresh):
         non_gt_inds = np.where(entry['gt_classes'] == 0)[0]
         if len(crowd_inds) == 0 or len(non_gt_inds) == 0:
             continue
-        iscrowd = [int(True) for _ in xrange(len(crowd_inds))]
+        iscrowd = [int(True) for _ in range(len(crowd_inds))]
         crowd_boxes = ds_utils.xyxy_to_xywh(entry['boxes'][crowd_inds, :])
         non_gt_boxes = ds_utils.xyxy_to_xywh(entry['boxes'][non_gt_inds, :])
         ious = COCOmask.iou(non_gt_boxes, crowd_boxes, iscrowd)
@@ -59,9 +59,9 @@ def __init__(self, image_set, year):
         self._COCO = COCO(self._get_ann_file())
         cats = self._COCO.loadCats(self._COCO.getCatIds())
         self._classes = tuple(['__background__'] + [c['name'] for c in cats])
-        self._class_to_ind = dict(zip(self.classes, xrange(self.num_classes)))
-        self._class_to_coco_cat_id = dict(zip([c['name'] for c in cats],
-                                              self._COCO.getCatIds()))
+        self._class_to_ind = dict(list(zip(self.classes, list(range(self.num_classes)))))
+        self._class_to_coco_cat_id = dict(list(zip([c['name'] for c in cats],
+                                              self._COCO.getCatIds())))
         self._image_index = self._load_image_set_index()
         # Default to roidb handler
         self.set_proposal_method('selective_search')
@@ -76,7 +76,7 @@ def __init__(self, image_set, year):
         }
         coco_name = image_set + year  # e.g., "val2014"
         self._data_name = (self._view_map[coco_name]
-                           if self._view_map.has_key(coco_name)
+                           if coco_name in self._view_map
                            else coco_name)
         # Dataset splits that have ground-truth annotations (test splits
         # do not have gt annotations)
@@ -140,9 +140,9 @@ def _roidb_from_proposals(self, method):
 
         if osp.exists(cache_file):
             with open(cache_file, 'rb') as fid:
-                roidb = cPickle.load(fid)
-            print '{:s} {:s} roidb loaded from {:s}'.format(self.name, method,
-                                                            cache_file)
+                roidb = pickle.load(fid)
+            print('{:s} {:s} roidb loaded from {:s}'.format(self.name, method,
+                                                            cache_file))
             return roidb
 
         if self._image_set in self._gt_splits:
@@ -154,8 +154,8 @@ def _roidb_from_proposals(self, method):
         else:
             roidb = self._load_proposals(method, None)
         with open(cache_file, 'wb') as fid:
-            cPickle.dump(roidb, fid, cPickle.HIGHEST_PROTOCOL)
-        print 'wrote {:s} roidb to {:s}'.format(method, cache_file)
+            pickle.dump(roidb, fid, pickle.HIGHEST_PROTOCOL)
+        print('wrote {:s} roidb to {:s}'.format(method, cache_file))
         return roidb
 
     def _load_proposals(self, method, gt_roidb):
@@ -177,10 +177,10 @@ def _load_proposals(self, method, gt_roidb):
             'edge_boxes_70']
         assert method in valid_methods
 
-        print 'Loading {} boxes'.format(method)
+        print('Loading {} boxes'.format(method))
         for i, index in enumerate(self._image_index):
             if i % 1000 == 0:
-                print '{:d} / {:d}'.format(i + 1, len(self._image_index))
+                print('{:d} / {:d}'.format(i + 1, len(self._image_index)))
 
             box_file = osp.join(
                 cfg.DATA_DIR, 'coco_proposals', method, 'mat',
@@ -213,16 +213,16 @@ def gt_roidb(self):
         cache_file = osp.join(self.cache_path, self.name + '_gt_roidb.pkl')
         if osp.exists(cache_file):
             with open(cache_file, 'rb') as fid:
-                roidb = cPickle.load(fid)
-            print '{} gt roidb loaded from {}'.format(self.name, cache_file)
+                roidb = pickle.load(fid)
+            print('{} gt roidb loaded from {}'.format(self.name, cache_file))
             return roidb
 
         gt_roidb = [self._load_coco_annotation(index)
                     for index in self._image_index]
 
         with open(cache_file, 'wb') as fid:
-            cPickle.dump(gt_roidb, fid, cPickle.HIGHEST_PROTOCOL)
-        print 'wrote gt roidb to {}'.format(cache_file)
+            pickle.dump(gt_roidb, fid, pickle.HIGHEST_PROTOCOL)
+        print('wrote gt roidb to {}'.format(cache_file))
         return gt_roidb
 
     def _load_coco_annotation(self, index):
@@ -306,18 +306,18 @@ def _get_thr_ind(coco_eval, thr):
         precision = \
             coco_eval.eval['precision'][ind_lo:(ind_hi + 1), :, :, 0, 2]
         ap_default = np.mean(precision[precision > -1])
-        print ('~~~~ Mean and per-category AP @ IoU=[{:.2f},{:.2f}] '
-               '~~~~').format(IoU_lo_thresh, IoU_hi_thresh)
-        print '{:.1f}'.format(100 * ap_default)
+        print(('~~~~ Mean and per-category AP @ IoU=[{:.2f},{:.2f}] '
+               '~~~~').format(IoU_lo_thresh, IoU_hi_thresh))
+        print('{:.1f}'.format(100 * ap_default))
         for cls_ind, cls in enumerate(self.classes):
             if cls == '__background__':
                 continue
             # minus 1 because of __background__
             precision = coco_eval.eval['precision'][ind_lo:(ind_hi + 1), :, cls_ind - 1, 0, 2]
             ap = np.mean(precision[precision > -1])
-            print '{:.1f}'.format(100 * ap)
+            print('{:.1f}'.format(100 * ap))
 
-        print '~~~~ Summary metrics ~~~~'
+        print('~~~~ Summary metrics ~~~~')
         coco_eval.summarize()
 
     def _do_detection_eval(self, res_file, output_dir):
@@ -330,8 +330,8 @@ def _do_detection_eval(self, res_file, output_dir):
         self._print_detection_eval_metrics(coco_eval)
         eval_file = osp.join(output_dir, 'detection_results.pkl')
         with open(eval_file, 'wb') as fid:
-            cPickle.dump(coco_eval, fid, cPickle.HIGHEST_PROTOCOL)
-        print 'Wrote COCO eval results to: {}'.format(eval_file)
+            pickle.dump(coco_eval, fid, pickle.HIGHEST_PROTOCOL)
+        print('Wrote COCO eval results to: {}'.format(eval_file))
 
     def _coco_results_one_category(self, boxes, cat_id):
         results = []
@@ -348,7 +348,7 @@ def _coco_results_one_category(self, boxes, cat_id):
               [{'image_id' : index,
                 'category_id' : cat_id,
                 'bbox' : [xs[k], ys[k], ws[k], hs[k]],
-                'score' : scores[k]} for k in xrange(dets.shape[0])])
+                'score' : scores[k]} for k in range(dets.shape[0])])
         return results
 
     def _write_coco_results_file(self, all_boxes, res_file):
@@ -360,12 +360,12 @@ def _write_coco_results_file(self, all_boxes, res_file):
         for cls_ind, cls in enumerate(self.classes):
             if cls == '__background__':
                 continue
-            print 'Collecting {} results ({:d}/{:d})'.format(cls, cls_ind,
-                                                          self.num_classes - 1)
+            print('Collecting {} results ({:d}/{:d})'.format(cls, cls_ind,
+                                                          self.num_classes - 1))
             coco_cat_id = self._class_to_coco_cat_id[cls]
             results.extend(self._coco_results_one_category(all_boxes[cls_ind],
                                                            coco_cat_id))
-        print 'Writing results json to {}'.format(res_file)
+        print('Writing results json to {}'.format(res_file))
         with open(res_file, 'w') as fid:
             json.dump(results, fid)
 

lib/datasets/factory.py

@@ -33,10 +33,10 @@
 
 def get_imdb(name):
     """Get an imdb (image database) by name."""
-    if not __sets.has_key(name):
+    if name not in __sets:
         raise KeyError('Unknown dataset: {}'.format(name))
     return __sets[name]()
 
 def list_imdbs():
     """List all registered imdbs."""
-    return __sets.keys()
+    return list(__sets.keys())