In the Github source code you provided，the test experiment is 45.86%.Can you check it again？

[Elaineok]

In the Github source code you provided，the test experiment is 45.86%.Can you check it again？

[vasgaowei]

In network.py, modify a line of code in return_ss_boxes function.
The original code is following:
def return_ss_boxes(self, boxes_index, mode='TRAIN'): if mode == 'TEST': return boxes_index box_num = min(500, len(boxes_index)) indexes = np.random.choice(boxes_index, size=box_num, replace=False) return indexes

The modified version is :
def return_ss_boxes(self, boxes_index, mode='TRAIN'): if mode == 'TEST': return boxes_index box_num = min(1000, len(boxes_index)) indexes = np.random.choice(boxes_index, size=box_num, replace=False) return indexes
You can even use all the proposals for training if you have enough GPU memory. Then you can set the number larger.