aromring
commented
3 years ago 
It's not a bug, it's a garbage signal (see above) which should be rejected as the code does, rightfully, at the min_autocorrelation_ratio = 0.5 level.
Now, while we are at it, it takes me quite some time to copy the raw data from postings, reformat it, add sample column, etc., - all for the sake of creating plots in Excel. Only to show you, for the n-th time, something you should have looked at yourself before posting. Thus, I have clarified point B in the "HOW TO REPORT BUGS" instructions: From now on, I will reject reports that don't contain plots of all relevant signals.
The calculated heart rate obtained by your algorithm is half of the actual heart rate, and the first local maximum of its relative autocorrelation is 0.29.
RAW IR DATA. 50Hz 3s 129650 129663 129675 129702 129688 129667 129666 129677 129648 129575 129485 129453 129461 129485 129506 129542 129565 129562 129583 129611 129624 129614 129602 129588 129614 129652 129672 129677 129679 129707 129710 129689 129708 129750 129763 129749 129751 129764 129780 129792 129782 129775 129782 129797 129782 129719 129640 129598 129596 129613 129615 129641 129657 129683 129696 129694 129703 129702 129693 129682 129668 129704 129719 129737 129735 129735 129752 129754 129756 129729 129746 129748 129759 129751 129752 129757 129750 129730 129723 129656 129583 129529 129511 129539 129537 129546 129574 129604 129624 129665 129690 129682 129685 129698 129712 129728 129743 129771 129781 129804 129785 129799 129805 129823 129839 129851 129864 129897 129902 129899 129881 129913 129920 129931 129927 129920 129927 129936 129941 129923 129885 129809 129766 129713 129730 129740 129763 129791 129819 129813 129817 129834 129842 129816 129810 129810 129806 129825 129846 129869 129909 129904 129905 129900 129917 129951 129973 129941 129959 129977