search

HelloRicky123 / Siamese-RPN

Full reimplementation of siamese rpn, has 0.24 eao on vot2017.
MIT License
223 stars 44 forks source link

I run your program and got only 0.1%AUC on OTB100 #48

Open Yvonneie opened 5 years ago

Yvonneie commented 5 years ago

Thanks for your work at first. I download your code and just change "data_path" in file "test_OTB100" to my folder of OTB100 sequences , and use the model "siamrpn_38"that you provided, but when I run "test_OTB100.py" ,I got only 0.1% AUC on that. and I don't know why. Is there anything I forgot, or anything I did wrong ?

Clannad000000 commented 5 years ago

Have you solved this problem?When I run the basketabll sequence,I find the results have many minus...

lgdhang commented 5 years ago

I have this problem either.

renlicheng commented 5 years ago

When I run "test_OTB100"，I had this problem In the test frame： Can you help me！！

/home/rlc/anaconda3/envs/H_R-Siamese-RPN/bin/python /home/rlc/H_R-Siamese-RPN/bin/test_OTB.py -ms /home/rlc/H_R-Siamese-RPN/imagenet_pretrain/siamrpn_38.pth -v tb100 0%| | 0/3 [00:00<?, ?it/s] 0%| | 0/98 [00:00<?, ?it/s]

0%| | 0/725 [00:00<?, ?it/s]

0%| | 1/725 [00:00<02:01, 5.96it/s]/home/rlc/H_R-Siamese-RPN/lib/utils.py:165: RuntimeWarning: overflow encountered in multiply box_w = anchor_w np.exp(offset_w) /home/rlc/H_R-Siamese-RPN/lib/utils.py:166: RuntimeWarning: overflow encountered in multiply box_h = anchor_h np.exp(offset_h) /home/rlc/H_R-Siamese-RPN/net/tracker.py:106: RuntimeWarning: overflow encountered in multiply sz2 = (w + pad) * (h + pad) /home/rlc/H_R-Siamese-RPN/net/tracker.py:102: RuntimeWarning: divide by zero encountered in true_divide return np.maximum(r, 1. / r) /home/rlc/H_R-Siamese-RPN/net/tracker.py:115: RuntimeWarning: divide by zero encountered in true_divide r_c = change((self.target_sz[0] / self.target_sz[1]) / (box_pred[:, 2] / box_pred[:, 3])) # ratio penalty /home/rlc/H_R-Siamese-RPN/net/tracker.py:115: RuntimeWarning: invalid value encountered in true_divide r_c = change((self.target_sz[0] / self.target_sz[1]) / (box_pred[:, 2] / box_pred[:, 3])) # ratio penalty 0%| | 1/725 [00:00<02:09, 5.58it/s] 0%| | 0/98 [00:02<?, ?it/s] 0%| | 0/3 [00:02<?, ?it/s] Traceback (most recent call last): File "/home/rlc/H_R-Siamese-RPN/bin/test_OTB.py", line 149, in result = run_SiamRPN(video_path, model_path, boxes[0]) File "/home/rlc/H_R-Siamese-RPN/net/run_SiamRPN.py", line 44, in run_SiamRPN bbox, score = tracker.update(frame) # x,y,w,h File "/home/rlc/H_R-Siamese-RPN/net/tracker.py", line 86, in update config.context_amount, self.img_mean) File "/home/rlc/H_R-Siamese-RPN/lib/utils.py", line 131, in get_instance_image instance_img, scale_x = crop_and_pad(img, cx, cy, size_x, s_x, img_mean) File "/home/rlc/H_R-Siamese-RPN/lib/utils.py", line 86, in crop_and_pad xmin = int(round_up(xmin + left)) ValueError: cannot convert float NaN to integer

chengxianda54321 commented 4 years ago

my model get 0.21 auc, f**k