cuda 10 compatibility

.gitignore

@@ -0,0 +1,15 @@
+output
+**/__pycache__
+build
+.idea
+**egg-info
+ic15_evaluate/ch4_gt
+datasets
+Res50_imagenet.pth
+maskrcnn_benchmark/*.so
+maskrcnn_benchmark.egg-info/SOURCES.txt
+ic15_evaluate/mb_ch4_results
+ic15_evaluate/all_detcors.txt
+ic15_evaluate/bo.json
+ic15_evaluate/evaluation_results.txt
+ic15_evaluate/rrc_evaluation_funcs.pyc

README.md

@@ -17,19 +17,19 @@ This is a pytorch-based implementation for paper [ContourNet](https://arxiv.org/
 
 ## Requirements
 
-We recommend you to use Anaconda [BaiduYun Link](https://pan.baidu.com/s/1_J9INU-UpiT43qormibAuw)(passward:1y3v) or [Geogle Drive](https://drive.google.com/file/d/1H64lTpR3xzlSRfUxfZa4dOhAYZJcO7RU/view?usp=sharing) to manage your libraries.
+We recommend you to use Anaconda [BaiduYun Link](https://pan.baidu.com/s/1_J9INU-UpiT43qormibAuw)(passward:1y3v) or [Google Drive](https://drive.google.com/file/d/1H64lTpR3xzlSRfUxfZa4dOhAYZJcO7RU/view?usp=sharing) to manage your libraries.
 
 
 ### Step-by-step install
 
 ```bash
-  conda create --name ContourNet
+  conda create --name ContourNet python=3.6
   conda activate ContourNet
   conda install ipython
   pip install ninja yacs cython matplotlib tqdm scipy shapely networkx pandas
-  conda install pytorch=1.0 torchvision=0.2 cudatoolkit=9.0 -c pytorch
+  conda install pytorch=1.0 torchvision=0.2 cudatoolkit=10.0 -c pytorch
   conda install -c menpo opencv
-  export INSTALL_DIR=$PWDcd $INSTALL_DIR
+  export INSTALL_DIR=$PWD
   git clone https://github.com/cocodataset/cocoapi.git
   cd cocoapi/PythonAPI
   python setup.py build_ext install
@@ -42,7 +42,7 @@ We recommend you to use Anaconda [BaiduYun Link](https://pan.baidu.com/s/1_J9INU
 ## Experiment on IC15 dataset
 ### Data preparing 
 #### step 1
-   Prepare data follow COCO format or you can download our [IC15dataset-BAIDU](https://pan.baidu.com/s/1GbF0PnWDKw3qn2o2XgpB7Q) (passward:ect5) or [Geogle Drive](https://drive.google.com/file/d/1ZWRQWJwhydoCsqdNlX80y94cKQedUywO/view?usp=sharing), and unzip it in 
+   Prepare data follow COCO format or you can download our [IC15dataset-BAIDU](https://pan.baidu.com/s/1GbF0PnWDKw3qn2o2XgpB7Q) (passward:ect5) or [Google Drive](https://drive.google.com/file/d/1ZWRQWJwhydoCsqdNlX80y94cKQedUywO/view?usp=sharing), and unzip it in 
 ```bash
    datasets/.
 ```

ic15_evaluate/eval_ic15.py

@@ -30,15 +30,15 @@
 ########## second step: evaluate results
 dirn =  "mb_ch4_results"
 eval_dir = "./"
-# lsc = [0.51+i*1.0/100 for i in range(30)]
-lsc = [0.65]
+lsc = list(range(20,90,1))
+# lsc = [0.65]
 
-f = open("evaluation_resutls.txt",'w')
+f = open("evaluation_results.txt",'w')
 
 fres = open('all_detcors.txt', 'r').readlines()
 
 for isc in lsc:
-	print 'Evaluating cf threshold {} ...'.format(str(isc)),
+	print('Evaluating cf threshold 0.{}:'.format(str(isc)))
 	sys.stdout.flush()
 
 	if os.path.exists("mb_ch4.zip"):
@@ -56,19 +56,19 @@
 		outName = os.path.join(dirn, filename)
 		with open(outName, 'a') as fout:
 			score = s[1].split(',')[-1].strip()
-			if float(score)<isc:
+			if float(score)<isc/100.:
 				continue
 			cors = ','.join(e for e in s[1].split(',')[:-1])
 			fout.writelines(cors+'\n')
 
 	os.chdir("mb_ch4_results/")
 	os.popen("zip -r ../mb_ch4.zip ./")
 	os.chdir("../")
-	cmd = "python "+eval_dir+"script.py "+"-g="+eval_dir+"ch4_gt.zip -s=mb_ch4.zip"
+	cmd = "python "+eval_dir+"script.py -g=ch4_gt.zip -s=mb_ch4.zip"
 	output = os.popen(cmd).read()
-	print(output)
-	f.writelines("========= "+str(isc)+":\n")
+	f.writelines("========= 0."+str(isc)+":\n")
 	lout = output.split('\n')
+	print("output is:",output)
 	f.writelines(lout[1]+'\n')
 	f.writelines(lout[2]+'\n')
 
@@ -78,6 +78,6 @@
 		shutil.rmtree("mb_ch4_results/")
 
 	f.flush()
-	time.sleep(0.01)
+	# time.sleep(0.01)
 
 f.close()

ic15_evaluate/rrc_evaluation_funcs.py

@@ -8,7 +8,6 @@
 import os
 import codecs
 import importlib
-from StringIO import StringIO
 
 def print_help():
     sys.stdout.write('Usage: python %s.py -g=<gtFile> -s=<submFile> [-o=<outputFolder> -p=<jsonParams>]' %sys.argv[0])
@@ -303,7 +302,7 @@ def main_evaluation(p,default_evaluation_params_fn,validate_data_fn,evaluate_met
         evalData = evaluate_method_fn(p['g'], p['s'], evalParams)
         resDict.update(evalData)
 
-    except Exception, e:
+    except Exception as e:
         resDict['Message']= str(e)
         resDict['calculated']=False
 
@@ -355,8 +354,8 @@ def main_validation(default_evaluation_params_fn,validate_data_fn):
             evalParams.update( p['p'] if isinstance(p['p'], dict) else json.loads(p['p'][1:-1]) )
 
         validate_data_fn(p['g'], p['s'], evalParams)              
-        print 'SUCCESS'
+        print('SUCCESS')
         sys.exit(0)
     except Exception as e:
-        print str(e)
+        print(str(e))
         sys.exit(101)

ic15_evaluate/script.py

@@ -1,4 +1,4 @@
-#!/usr/bin/env pytho
+#!/usr/bin/env python
 # -*- coding: utf-8 -*-
 from collections import namedtuple
 import rrc_evaluation_funcs

maskrcnn_benchmark/data/datasets/evaluation/word/word_eval.py

@@ -244,7 +244,7 @@ def boundary_to_mask(bo_x, bo_y, name, num):
 def mask_to_roRect(mask, img_shape):
     ## convert mask into rotated rect
     e = mask[0, :, :]
-    _, countours, hier = cv2.findContours(e.clone().numpy(), cv2.RETR_CCOMP, cv2.CHAIN_APPROX_NONE)  # Aarlog
+    countours, hier = cv2.findContours(e.clone().numpy(), cv2.RETR_CCOMP, cv2.CHAIN_APPROX_NONE)  # Aarlog
     if len(countours) == 0:
         return np.zeros((1, 8))
     t_c = countours[0].copy()

requirements.txt

@@ -3,3 +3,7 @@ yacs
 cython
 matplotlib
 tqdm
+scipy
+shapely
+networkx
+pandas