Compute_Hr function miscalculates.

I am working on the MIT-BIH records. The aim is to calculate the Heart Rate and find the maximum, minimum and the mean of the Heart rates. I extracted the ECG signal from the record, calculated the QRS peaks and calculated the Heart Rates using the compute_hr function from the wfdb package. Similarly extracted the QRS peaks from the annotation file and calculated the Heart Rates using the compute_hr function from the wfdb package. For the 101 record, the maximum Heart Rate from the annotations is 900.00bpm while the calculated qrs indices gave a maximum Heart rate of 111.3402bpm. I have this issue with many other records.

Python code:

record = wfdb.rdrecord(record_path, channels=[0])                                                                                                                                                                 
ann_ref = wfdb.rdann(record_path,'atr')
qrs_inds = processing.gqrs_detect(sig=record.p_signal[:,0], fs=record.fs)

heart_rate_wfdb = processing.compute_hr(sig_len=record.p_signal.shape[0], qrs_inds=qrs_inds, fs=record.fs)
heart_rate_wfdb = heart_rate_wfdb[np.logical_not(np.isnan(heart_rate_wfdb))]
heart_rate_min = "{0:.4f}".format(np.amin(heart_rate_wfdb))
heart_rate_max = "{0:.4f}".format(np.amax(heart_rate_wfdb))
heart_rate_mean = "{0:.4f}".format(np.mean(heart_rate_wfdb))

ref_heart_rate_wfdb = processing.compute_hr(sig_len=record.p_signal.shape[0], qrs_inds=ann_ref.sample[1:], fs=record.fs)
ref_heart_rate_wfdb = ref_heart_rate_wfdb[np.logical_not(np.isnan(ref_heart_rate_wfdb))]
ref_heart_rate_min = "{0:.4f}".format(np.amin(ref_heart_rate_wfdb))
ref_heart_rate_max = "{0:.4f}".format(np.amax(ref_heart_rate_wfdb))
ref_heart_rate_mean = "{0:.4f}".format(np.mean(ref_heart_rate_wfdb))`

Can someone tell me what is that I am doing wrong?

Start at the processing.compute_hr entry point and reproduce the comparison using record 101, qrs_inds, and ann_ref.sample[1:] from the issue. Check how the two QRS index inputs produce different heart-rate extrema; done means the discrepancy is explained and any confirmed defect is covered by a regression test.