training is normal, detection is bad.

[tableter]

first think you for you great job. I got a problem when I try to training my own dataset, this is my loss when training, I think it is looks fine.

counter:[ 0], loss:33.13054276, loc_loss:6.43029118, conf_loss:26.70025063, conf_pos_loss:0.00142080, conf_neg_loss:26.69882965, reg_loss:1.38765657 counter:[100], loss:4.54007387, loc_loss:1.85214794, conf_loss:2.68792582, conf_pos_loss:2.00233412, conf_neg_loss:0.68559170, reg_loss:1.50877857 counter:[200], loss:6.03843784, loc_loss:3.32742476, conf_loss:2.71101284, conf_pos_loss:1.22818446, conf_neg_loss:1.48282850, reg_loss:1.50922894 counter:[300], loss:3.69741201, loc_loss:1.14273107, conf_loss:2.55468082, conf_pos_loss:1.85684276, conf_neg_loss:0.69783795, reg_loss:1.51035881 counter:[400], loss:6.67551851, loc_loss:3.63578224, conf_loss:3.03973627, conf_pos_loss:1.45538306, conf_neg_loss:1.58435321, reg_loss:1.51230383 counter:[500], loss:4.25586843, loc_loss:1.50796425, conf_loss:2.74790430, conf_pos_loss:1.70701385, conf_neg_loss:1.04089057, reg_loss:1.51697719 counter:[600], loss:4.75161076, loc_loss:2.01842546, conf_loss:2.73318553, conf_pos_loss:1.57193756, conf_neg_loss:1.16124797, reg_loss:1.52445817 counter:[700], loss:3.82058096, loc_loss:1.37005222, conf_loss:2.45052886, conf_pos_loss:1.49491525, conf_neg_loss:0.95561361, reg_loss:1.54858720 counter:[800], loss:3.19424558, loc_loss:0.87267661, conf_loss:2.32156897, conf_pos_loss:1.38372529, conf_neg_loss:0.93784380, reg_loss:1.57513905 counter:[900], loss:3.19253016, loc_loss:0.82195914, conf_loss:2.37057090, conf_pos_loss:1.61763763, conf_neg_loss:0.75293320, reg_loss:1.61062586 counter:[1000], loss:3.07513142, loc_loss:0.80792892, conf_loss:2.26720238, conf_pos_loss:1.37672997, conf_neg_loss:0.89047241, reg_loss:1.64051330 counter:[1100], loss:2.63166571, loc_loss:0.42047766, conf_loss:2.21118808, conf_pos_loss:1.44011438, conf_neg_loss:0.77107370, reg_loss:1.66668820 counter:[1200], loss:2.42649984, loc_loss:0.27569529, conf_loss:2.15080452, conf_pos_loss:1.43105984, conf_neg_loss:0.71974456, reg_loss:1.69701743 counter:[1300], loss:2.18804979, loc_loss:0.33698606, conf_loss:1.85106385, conf_pos_loss:1.03611267, conf_neg_loss:0.81495118, reg_loss:1.73790741 counter:[1400], loss:1.30303633, loc_loss:0.18249714, conf_loss:1.12053919, conf_pos_loss:0.71575379, conf_neg_loss:0.40478539, reg_loss:1.79356992 counter:[1500], loss:1.06385875, loc_loss:0.27102375, conf_loss:0.79283506, conf_pos_loss:0.51907861, conf_neg_loss:0.27375644, reg_loss:1.85341001 counter:[1600], loss:0.85193586, loc_loss:0.32023808, conf_loss:0.53169775, conf_pos_loss:0.27517420, conf_neg_loss:0.25652358, reg_loss:1.90626478 counter:[1700], loss:0.34444326, loc_loss:0.19898847, conf_loss:0.14545481, conf_pos_loss:0.09046964, conf_neg_loss:0.05498517, reg_loss:1.94898546 counter:[1800], loss:0.32737771, loc_loss:0.21614154, conf_loss:0.11123617, conf_pos_loss:0.05481290, conf_neg_loss:0.05642327, reg_loss:1.97096920 counter:[1900], loss:0.20833418, loc_loss:0.16160801, conf_loss:0.04672617, conf_pos_loss:0.02719803, conf_neg_loss:0.01952814, reg_loss:1.98403728 counter:[2000], loss:0.13615118, loc_loss:0.09437951, conf_loss:0.04177167, conf_pos_loss:0.02639612, conf_neg_loss:0.01537555, reg_loss:1.99416208 counter:[2100], loss:0.11877032, loc_loss:0.08130707, conf_loss:0.03746326, conf_pos_loss:0.01944295, conf_neg_loss:0.01802031, reg_loss:2.00119495 counter:[2200], loss:0.22725815, loc_loss:0.18989232, conf_loss:0.03736583, conf_pos_loss:0.01977556, conf_neg_loss:0.01759027, reg_loss:2.00774050 counter:[2300], loss:0.14290509, loc_loss:0.11903250, conf_loss:0.02387259, conf_pos_loss:0.00866592, conf_neg_loss:0.01520667, reg_loss:2.01375747 counter:[2400], loss:0.09723526, loc_loss:0.07686017, conf_loss:0.02037509, conf_pos_loss:0.00705688, conf_neg_loss:0.01331820, reg_loss:2.02066350 counter:[2500], loss:0.08595456, loc_loss:0.07271308, conf_loss:0.01324149, conf_pos_loss:0.00738176, conf_neg_loss:0.00585973, reg_loss:2.02684069 counter:[2600], loss:0.05133056, loc_loss:0.04061699, conf_loss:0.01071357, conf_pos_loss:0.00543802, conf_neg_loss:0.00527555, reg_loss:2.03239059 counter:[2700], loss:0.09128784, loc_loss:0.07061297, conf_loss:0.02067486, conf_pos_loss:0.00962903, conf_neg_loss:0.01104584, reg_loss:2.03830624 counter:[2800], loss:0.06412772, loc_loss:0.05351655, conf_loss:0.01061117, conf_pos_loss:0.00606804, conf_neg_loss:0.00454313, reg_loss:2.04460669 counter:[2900], loss:0.17229854, loc_loss:0.15321457, conf_loss:0.01908397, conf_pos_loss:0.00611297, conf_neg_loss:0.01297100, reg_loss:2.05361724 counter:[3000], loss:0.05515540, loc_loss:0.04634700, conf_loss:0.00880840, conf_pos_loss:0.00422499, conf_neg_loss:0.00458341, reg_loss:2.05993342 counter:[3100], loss:0.15578209, loc_loss:0.14866345, conf_loss:0.00711864, conf_pos_loss:0.00341933, conf_neg_loss:0.00369931, reg_loss:2.06578898 counter:[3200], loss:0.06318847, loc_loss:0.04772605, conf_loss:0.01546242, conf_pos_loss:0.00874956, conf_neg_loss:0.00671285, reg_loss:2.07291412 counter:[3300], loss:0.05998845, loc_loss:0.04955832, conf_loss:0.01043014, conf_pos_loss:0.00463177, conf_neg_loss:0.00579837, reg_loss:2.07992125 counter:[3400], loss:0.06479763, loc_loss:0.05734175, conf_loss:0.00745588, conf_pos_loss:0.00237658, conf_neg_loss:0.00507929, reg_loss:2.08763075 counter:[3500], loss:0.06136370, loc_loss:0.05244459, conf_loss:0.00891911, conf_pos_loss:0.00415542, conf_neg_loss:0.00476369, reg_loss:2.09559011 counter:[3600], loss:0.05930467, loc_loss:0.04914257, conf_loss:0.01016210, conf_pos_loss:0.00614242, conf_neg_loss:0.00401968, reg_loss:2.10123897 counter:[3700], loss:0.16376168, loc_loss:0.15444940, conf_loss:0.00931227, conf_pos_loss:0.00487474, conf_neg_loss:0.00443753, reg_loss:2.10925007 counter:[3800], loss:0.16854750, loc_loss:0.15668774, conf_loss:0.01185976, conf_pos_loss:0.00543092, conf_neg_loss:0.00642883, reg_loss:2.11939335 counter:[3900], loss:0.06995875, loc_loss:0.05945601, conf_loss:0.01050274, conf_pos_loss:0.00473152, conf_neg_loss:0.00577122, reg_loss:2.13219547 counter:[4000], loss:0.05760749, loc_loss:0.04959865, conf_loss:0.00800884, conf_pos_loss:0.00201904, conf_neg_loss:0.00598980, reg_loss:2.14414692 counter:[4100], loss:0.03488719, loc_loss:0.02875672, conf_loss:0.00613047, conf_pos_loss:0.00270938, conf_neg_loss:0.00342109, reg_loss:2.15656161 counter:[4200], loss:0.03834939, loc_loss:0.03179526, conf_loss:0.00655413, conf_pos_loss:0.00346161, conf_neg_loss:0.00309251, reg_loss:2.16395712 counter:[4300], loss:0.02187771, loc_loss:0.01646716, conf_loss:0.00541055, conf_pos_loss:0.00217170, conf_neg_loss:0.00323886, reg_loss:2.16975951 counter:[4400], loss:0.08669566, loc_loss:0.07352066, conf_loss:0.01317500, conf_pos_loss:0.00545273, conf_neg_loss:0.00772227, reg_loss:2.18105173 counter:[4500], loss:0.07833251, loc_loss:0.07153510, conf_loss:0.00679742, conf_pos_loss:0.00130077, conf_neg_loss:0.00549665, reg_loss:2.19855976 counter:[4600], loss:1.26507616, loc_loss:0.50168902, conf_loss:0.76338714, conf_pos_loss:0.43574136, conf_neg_loss:0.32764578, reg_loss:2.37931824 counter:[4700], loss:0.19005077, loc_loss:0.13030116, conf_loss:0.05974960, conf_pos_loss:0.01502002, conf_neg_loss:0.04472958, reg_loss:2.56408429 counter:[4800], loss:0.07824517, loc_loss:0.06682102, conf_loss:0.01142415, conf_pos_loss:0.00373227, conf_neg_loss:0.00769188, reg_loss:2.58001375 counter:[4900], loss:0.02106231, loc_loss:0.01282067, conf_loss:0.00824164, conf_pos_loss:0.00443847, conf_neg_loss:0.00380317, reg_loss:2.58456993 counter:[5000], loss:0.03256434, loc_loss:0.02461102, conf_loss:0.00795332, conf_pos_loss:0.00506878, conf_neg_loss:0.00288454, reg_loss:2.58760643

but when I try to detect, I got nothing, I found that all conf_pred<TEST_SCORE_THRESHOLD. I do not know why, maybe that I did not chage the PRIOR_WIDTHS and PRIOR_HEIGHTS cause this problem? Besides, I can not download your paper ,can you send it to my email? qiaoyan272180@163.com. thank you very much.