原来运行demo.py是成功的。训练后对OTB100评估结果如下

|Tracker name| Success | Norm Precision | Precision |

| SiamRPN++ | 0.609 | 0.000 | 0.808 |

训练过后 python demo.py --config ../experiments/siamrpn_r50_l234_dwxcorr/config.yaml --snapshot ../experiments/siamrpn_r50_l234_dwxcorr/snapshot/checkpoint_e20.pth --video ../demo/bag.avi
可以打开窗口并画框，开始追踪发生如下警告
C:\Users/---/Desktop/pysot-master\pysot\tracker\siamrpn_tracker.py:53: RuntimeWarning: overflow encountered in exp
delta[2, :] = np.exp(delta[2, :]) * anchor[:, 2]
C:\Users/---/Desktop/pysot-master\pysot\tracker\siamrpn_tracker.py:54: RuntimeWarning: overflow encountered in exp
delta[3, :] = np.exp(delta[3, :]) * anchor[:, 3]
C:\Users/---/Desktop/pysot-master\pysot\tracker\siamrpn_tracker.py:115: RuntimeWarning: divide by zero encountered in divide
return np.maximum(r, 1. / r)
C:\Users/---/Desktop/pysot-master\pysot\tracker\siamrpn_tracker.py:127: RuntimeWarning: invalid value encountered in divide
(pred_bbox[2, :]/pred_bbox[3, :]))
C:\Users/---/Desktop/pysot-master\pysot\tracker\siamrpn_tracker.py:126: RuntimeWarning: divide by zero encountered in divide
r_c = change((self.size[0]/self.size[1]) /
C:\Users/---/Desktop/pysot-master\pysot\tracker\siamrpn_tracker.py:127: RuntimeWarning: divide by zero encountered in divide
(pred_bbox[2, :]/pred_bbox[3, :]))
C:\Users/---/Desktop/pysot-master\pysot\tracker\siamrpn_tracker.py:53: RuntimeWarning: overflow encountered in multiply
delta[2, :] = np.exp(delta[2, :]) * anchor[:, 2]
.....
运行时窗口内角落上有绿色的一点

使用model_zoo中的siamrpn_r50_l234_dwxcorr model 可以成功运行

使用训练过的模型时，#320按这个改过。