Testing Code for MultiProposal

test.py

import mxnet as mx
from mxnet import nd
import numpy as np
import time

np.random.seed(1234)

feature_stride = 16 
scales = (8, 16, 32)
ratios = (0.5, 1, 2)
rpn_pre_nms_top_n = 12000 
rpn_post_nms_top_n = 2000
threshold = 0.5
rpn_min_size = 16

batch_size = 16 
feat_len = 128
H, W = feat_len, feat_len
num_anchors = len(scales) * len(ratios)
count_anchors = feat_len * feat_len * num_anchors

T = 0.0

'''
    cls_prob: (batch_size, 2 * num_anchors, H, W)
    bbox_pred: (batch_size, 4 * num_anchors, H, W) 
    im_info: (batch_size, 3)
'''

ctx = mx.cpu(0)
im_info = nd.empty((batch_size, 3), dtype = np.float32,ctx = ctx) 

cls_prob = nd.random.uniform(shape = (batch_size, 2 * num_anchors, H, W), ctx = ctx)
bbox_pred = nd.random.uniform(shape = (batch_size, 4 * num_anchors, H, W), ctx = ctx)

print ("start")
for i in range(batch_size):
    im_len = feat_len * feature_stride
    im_size = np.array([im_len, im_len]) #np.random.randint(100, feat_len * feature_stride, size = (2,))
    im_scale = np.random.randint(70, 100) / 100.0
    im_info[i, :] = [im_size[0], im_size[1], im_scale]

def test(rpn_pre_nms_top_n, rpn_post_nms_top_n):
    oldtime = time.time()
    multi_proposal, multi_score = mx.nd.contrib.MultiProposal(
            cls_score = cls_prob,
            bbox_pred = bbox_pred,
            im_info = im_info,
            feature_stride = feature_stride,
            scales = scales,
            ratios = ratios,
            rpn_pre_nms_top_n = rpn_pre_nms_top_n,
            rpn_post_nms_top_n = rpn_post_nms_top_n,
            threshold = threshold,
            rpn_min_size = rpn_min_size, output_score = True)
    nd.waitall()
    dt = time.time() - oldtime
    global T
    T += dt

print ("Testing...")
lastT = T
for i in range(10):
    test(rpn_pre_nms_top_n, rpn_post_nms_top_n)
    test(rpn_pre_nms_top_n, 1500)
    test(1000, 500)
    print (T, T - lastT)
    lastT = T
print ("test ok")
print (T)