Add files via upload

comparisons_wacv.py

@@ -0,0 +1,105 @@
+"""
+Bar chart demo with pairs of bars grouped for easy comparison.
+"""
+import numpy as np
+import matplotlib.pyplot as plt
+from matplotlib.ticker import MaxNLocator
+
+# n_groups = 3
+
+
+
+# plt.bar(range(len(data)), data)
+# plt.show()
+#plt.xlabel('Data Imbalance')
+#plt.ylabel('Accuracy')
+
+# plt.title('Comparison Between Different Data Imbalance Rate')
+# #plt.legend()
+#
+
+# plt.show()
+#plt.savefig(u't.png')
+# fig, axes = plt.subplots(2,2)
+# axes.yaxis.set_major_locator(MaxNLocator(prune='upper'))
+# plt.tick_params(labelcolor='none', top='off', bottom='off', left='off', right='off')
+
+fig, ((ax1, ax2, ax3), (ax4, ax5, ax6)) = plt.subplots(nrows=2,ncols=3,figsize=(10,5))
+
+data = [67.1, 67.3, 66.9]
+index = np.arange(len(data))
+ax1.bar(index, data, width = 0.3,color = 'b')
+ax1.set_xticks(index)
+ax1.set_xticklabels(('1:1', '1:2', '1:3'))
+ax1.set_ylim(66.5, 67.5)
+ax1.tick_params(axis='y', colors='black')
+ax1.set_ylabel('accuracy',color = 'black')
+# plt.xlabel('imbalance')
+ax1.set_title('(a) Imbalance', y = -0.3)
+
+# plt.subplots_adjust(right = 0.1)
+data = [66.7,67.3,67.1,66.8,66.6]
+index = np.arange(len(data))
+ax2.bar(index, data, width = 0.3,color = 'b')
+ax2.set_xticks(index)
+ax2.set_xticklabels(('9','18','36','72','108'))
+ax2.set_ylim(66.5, 67.5)
+ax2.tick_params(axis='y', colors='black')
+ax2.set_title('(b) Batch Size', y = -0.3)
+ax2.set_ylabel('accuracy',color = 'black')
+
+data = [65.9,66.5,67.3,67.3]
+index = np.arange(len(data))
+ax3.bar(index, data, width = 0.3,color = 'b')
+ax3.set_xticks(index)
+ax3.set_xticklabels(('1-gram', '2-gram', '3-gram', '4-gram'))
+ax3.set_ylim(65, 68)
+ax3.tick_params(axis='y', colors='black')
+ax3.set_ylabel('accuracy',color= 'black')
+ax3.set_title('(c) Convolutional N-gram', y = -0.3)
+
+
+data1 = [66.3, 52.8, 50.7, 67.3]
+data2 = [50 + (i - 2.5) / 12.5 * 18.0 for i in [14,3,4,9]]
+print data2
+index = np.arange(len(data1))
+tax = ax4.twinx()
+
+ax4.bar(index - 0.15, data1, width = 0.3,color = 'b')
+ax4.bar(index + 0.15, data2, width = 0.3,color = 'black')
+ax4.set_xticks(index)
+ax4.set_xticklabels(('No BN','Data', 'QI', 'QIA'))
+ax4.set_ylim(50, 68)
+ax4.set_ylabel('accuracy',color = 'black')
+ax4.set_title('(d) Batch Normalization', y = -0.3)
+ax4.tick_params(axis='y', colors='black')
+tax.set_ylabel('epoch',color = 'black')
+tax.set_ylim(2,16)
+tax.tick_params(axis='y', colors='black')
+
+data = [67.3,67.5,67.4,67.2,67.0]
+index = np.arange(len(data))
+ax5.bar(index, data, width = 0.3,color = 'b')
+ax5.set_xticks(index)
+ax5.set_xticklabels(('0.1','0.2','0.3','0.4','0.5'))
+ax5.set_ylim(66.7, 67.8)
+ax5.tick_params(axis='y', colors='black')
+ax5.set_title('(e) lambda2', y = -0.3)
+ax5.set_ylabel('accuracy',color = 'black')
+
+
+data = [67.3,67.5,67.4]
+index = np.arange(len(data))
+ax6.bar(index, data, width = 0.3,color = 'b')
+ax6.set_xticks(index)
+ax6.set_xticklabels(('0.1','0.2','0.3'))
+ax6.set_ylim(67.2, 67.6)
+ax6.tick_params(axis='y', colors='black')
+ax6.set_title('(f) margin', y = -0.3)
+ax6.set_ylabel('accuracy',color = 'black')
+
+
+
+plt.tight_layout()
+# plt.show()
+plt.savefig('comparisons_wacv.eps',format = 'eps',dpi = 1000)