calculation mAP (mean average precision)...
6000
 detections_count = 9091, unique_truth_count = 6000  
class_id = 0, name = a, ap = 100.00%   	 (TP = 216, FP = 1) 
class_id = 1, name = space, ap = 98.59%   	 (TP = 218, FP = 6) 
class_id = 2, name = b, ap = 100.00%   	 (TP = 203, FP = 0) 
class_id = 3, name = c, ap = 99.99%   	 (TP = 227, FP = 0) 
class_id = 4, name = d, ap = 100.00%   	 (TP = 223, FP = 0) 
class_id = 5, name = e, ap = 86.90%   	 (TP = 129, FP = 26) 
class_id = 6, name = f, ap = 100.00%   	 (TP = 225, FP = 1) 
class_id = 7, name = g, ap = 100.00%   	 (TP = 218, FP = 0) 
class_id = 8, name = h, ap = 100.00%   	 (TP = 252, FP = 1) 
class_id = 9, name = i, ap = 98.82%   	 (TP = 196, FP = 2) 
class_id = 10, name = j, ap = 100.00%   	 (TP = 217, FP = 0) 
class_id = 11, name = k, ap = 100.00%   	 (TP = 243, FP = 0) 
class_id = 12, name = l, ap = 100.00%   	 (TP = 222, FP = 1) 
class_id = 13, name = m, ap = 100.00%   	 (TP = 229, FP = 0) 
class_id = 14, name = n, ap = 93.93%   	 (TP = 200, FP = 88) 
class_id = 15, name = o, ap = 100.00%   	 (TP = 217, FP = 4) 
class_id = 16, name = p, ap = 100.00%   	 (TP = 241, FP = 16) 
class_id = 17, name = q, ap = 100.00%   	 (TP = 235, FP = 15) 
class_id = 18, name = r, ap = 99.52%   	 (TP = 184, FP = 0) 
class_id = 19, name = s, ap = 100.00%   	 (TP = 222, FP = 2) 
class_id = 20, name = t, ap = 99.98%   	 (TP = 212, FP = 24) 
class_id = 21, name = u, ap = 100.00%   	 (TP = 198, FP = 1) 
class_id = 22, name = v, ap = 74.66%   	 (TP = 76, FP = 0) 
class_id = 23, name = w, ap = 100.00%   	 (TP = 205, FP = 0) 
class_id = 24, name = x, ap = 100.00%   	 (TP = 245, FP = 0) 
class_id = 25, name = y, ap = 100.00%   	 (TP = 223, FP = 0) 
class_id = 26, name = z, ap = 100.00%   	 (TP = 223, FP = 0) 

 for thresh = 0.25, precision = 0.97, recall = 0.95, F1-score = 0.96 
 for thresh = 0.25, TP = 5699, FP = 188, FN = 301, average IoU = 87.69 % 
 IoU threshold = 50 %, used Area-Under-Curve for each unique Recall 
 mean average precision (mAP@0.50) = 0.982366, or 98.24 % 
 mean_average_precision (mAP@0.5) = 0.982366