Initial working version
This commit is contained in:
hd committed
1 parent
40a6fe1b5f
commit
2f726ca795
1 file changed
+70
-71
@@ -464,9 +464,6 @@ class ConfusionMatrix:
|
||||
return fig
|
||||
|
||||
|
||||
import torch
|
||||
|
||||
|
||||
@dataclass(frozen=True)
|
||||
class MeanAveragePrecision:
|
||||
map: float
|
||||
@@ -572,7 +569,7 @@ class MeanAveragePrecision:
|
||||
stats = [np.concatenate(x, 0) for x in zip(*stats)]
|
||||
|
||||
if len(stats) and stats[0].any():
|
||||
tp, fp, p, r, f1, ap, ap_class = ap_per_class(*stats, names=classes)
|
||||
tp, fp, p, r, f1, ap, ap_class = MeanAveragePrecision.ap_per_class(*stats, names=classes)
|
||||
ap50, ap = ap[:, 0], ap.mean(1) # AP@0.5, AP@0.5:0.95
|
||||
mp, mr, map50, map = p.mean(), r.mean(), ap50.mean(), ap.mean()
|
||||
|
||||
@@ -633,82 +630,84 @@ class MeanAveragePrecision:
|
||||
matches = matches[np.unique(matches[:, 0], return_index=True)[1]]
|
||||
return matches
|
||||
|
||||
@staticmethod
|
||||
def compute_ap(recall, precision):
|
||||
"""Compute the average precision, given the recall and precision curves
|
||||
# Arguments
|
||||
recall: The recall curve (list)
|
||||
precision: The precision curve (list)
|
||||
# Returns
|
||||
Average precision, precision curve, recall curve
|
||||
"""
|
||||
# Append sentinel values to beginning and end
|
||||
mrec = np.concatenate(([0.0], recall, [1.0]))
|
||||
mpre = np.concatenate(([1.0], precision, [0.0]))
|
||||
|
||||
def ap_per_class(tp, conf, pred_cls, target_cls, eps=1e-16):
|
||||
i = np.argsort(-conf)
|
||||
tp, conf, pred_cls = tp[i], conf[i], pred_cls[i]
|
||||
# Compute the precision envelope
|
||||
mpre = np.flip(np.maximum.accumulate(np.flip(mpre)))
|
||||
|
||||
unique_classes, nt = np.unique(target_cls, return_counts=True)
|
||||
nc = unique_classes.shape[0] # number of classes, number of detections
|
||||
# Integrate area under curve
|
||||
method = "interp" # methods: 'continuous', 'interp'
|
||||
if method == "interp":
|
||||
x = np.linspace(0, 1, 101) # 101-point interp (COCO)
|
||||
ap = np.trapz(np.interp(x, mrec, mpre), x) # integrate
|
||||
else: # 'continuous'
|
||||
i = np.where(mrec[1:] != mrec[:-1])[0] # points where x axis (recall) changes
|
||||
ap = np.sum((mrec[i + 1] - mrec[i]) * mpre[i + 1]) # area under curve
|
||||
|
||||
px, py = np.linspace(0, 1, 1000), [] # for plotting
|
||||
ap, p, r = np.zeros((nc, tp.shape[1])), np.zeros((nc, 1000)), np.zeros((nc, 1000))
|
||||
for ci, c in enumerate(unique_classes):
|
||||
i = pred_cls == c
|
||||
n_l = nt[ci] # number of labels
|
||||
n_p = i.sum() # number of predictions
|
||||
if n_p == 0 or n_l == 0:
|
||||
continue
|
||||
return ap, mpre, mrec
|
||||
|
||||
# Accumulate FPs and TPs
|
||||
fpc = (1 - tp[i]).cumsum(0)
|
||||
tpc = tp[i].cumsum(0)
|
||||
@staticmethod
|
||||
def ap_per_class(tp, conf, pred_cls, target_cls, eps=1e-16):
|
||||
i = np.argsort(-conf)
|
||||
tp, conf, pred_cls = tp[i], conf[i], pred_cls[i]
|
||||
|
||||
# Recall
|
||||
recall = tpc / (n_l + eps) # recall curve
|
||||
r[ci] = np.interp(
|
||||
-px, -conf[i], recall[:, 0], left=0
|
||||
) # negative x, xp because xp decreases
|
||||
unique_classes, nt = np.unique(target_cls, return_counts=True)
|
||||
nc = unique_classes.shape[0] # number of classes, number of detections
|
||||
|
||||
# Precision
|
||||
precision = tpc / (tpc + fpc) # precision curve
|
||||
p[ci] = np.interp(-px, -conf[i], precision[:, 0], left=1) # p at pr_score
|
||||
px, py = np.linspace(0, 1, 1000), [] # for plotting
|
||||
ap, p, r = np.zeros((nc, tp.shape[1])), np.zeros((nc, 1000)), np.zeros((nc, 1000))
|
||||
for ci, c in enumerate(unique_classes):
|
||||
i = pred_cls == c
|
||||
n_l = nt[ci] # number of labels
|
||||
n_p = i.sum() # number of predictions
|
||||
if n_p == 0 or n_l == 0:
|
||||
continue
|
||||
|
||||
# AP from recall-precision curve
|
||||
for j in range(tp.shape[1]):
|
||||
ap[ci, j], mpre, mrec = compute_ap(recall[:, j], precision[:, j])
|
||||
# Accumulate FPs and TPs
|
||||
fpc = (1 - tp[i]).cumsum(0)
|
||||
tpc = tp[i].cumsum(0)
|
||||
|
||||
# Compute F1 (harmonic mean of precision and recall)
|
||||
f1 = 2 * p * r / (p + r + eps)
|
||||
# Recall
|
||||
recall = tpc / (n_l + eps) # recall curve
|
||||
r[ci] = np.interp(
|
||||
-px, -conf[i], recall[:, 0], left=0
|
||||
) # negative x, xp because xp decreases
|
||||
|
||||
i = smooth(f1.mean(0), 0.1).argmax() # max F1 index
|
||||
p, r, f1 = p[:, i], r[:, i], f1[:, i]
|
||||
tp = (r * nt).round() # true positives
|
||||
fp = (tp / (p + eps) - tp).round() # false positives
|
||||
return tp, fp, p, r, f1, ap, unique_classes.astype(int)
|
||||
# Precision
|
||||
precision = tpc / (tpc + fpc) # precision curve
|
||||
p[ci] = np.interp(-px, -conf[i], precision[:, 0], left=1) # p at pr_score
|
||||
|
||||
# AP from recall-precision curve
|
||||
for j in range(tp.shape[1]):
|
||||
ap[ci, j], mpre, mrec = MeanAveragePrecision.compute_ap(recall[:, j], precision[:, j])
|
||||
|
||||
# Compute F1 (harmonic mean of precision and recall)
|
||||
f1 = 2 * p * r / (p + r + eps)
|
||||
|
||||
i = MeanAveragePrecision.smooth(f1.mean(0), 0.1).argmax() # max F1 index
|
||||
p, r, f1 = p[:, i], r[:, i], f1[:, i]
|
||||
tp = (r * nt).round() # true positives
|
||||
fp = (tp / (p + eps) - tp).round() # false positives
|
||||
return tp, fp, p, r, f1, ap, unique_classes.astype(int)
|
||||
|
||||
@staticmethod
|
||||
def smooth(y, f=0.05):
|
||||
# Box filter of fraction f
|
||||
nf = round(len(y) * f * 2) // 2 + 1 # number of filter elements (must be odd)
|
||||
p = np.ones(nf // 2) # ones padding
|
||||
yp = np.concatenate((p * y[0], y, p * y[-1]), 0) # y padded
|
||||
return np.convolve(yp, np.ones(nf) / nf, mode="valid") # y-smoothed
|
||||
|
||||
|
||||
def smooth(y, f=0.05):
|
||||
# Box filter of fraction f
|
||||
nf = round(len(y) * f * 2) // 2 + 1 # number of filter elements (must be odd)
|
||||
p = np.ones(nf // 2) # ones padding
|
||||
yp = np.concatenate((p * y[0], y, p * y[-1]), 0) # y padded
|
||||
return np.convolve(yp, np.ones(nf) / nf, mode="valid") # y-smoothed
|
||||
|
||||
|
||||
def compute_ap(recall, precision):
|
||||
"""Compute the average precision, given the recall and precision curves
|
||||
# Arguments
|
||||
recall: The recall curve (list)
|
||||
precision: The precision curve (list)
|
||||
# Returns
|
||||
Average precision, precision curve, recall curve
|
||||
"""
|
||||
|
||||
# Append sentinel values to beginning and end
|
||||
mrec = np.concatenate(([0.0], recall, [1.0]))
|
||||
mpre = np.concatenate(([1.0], precision, [0.0]))
|
||||
|
||||
# Compute the precision envelope
|
||||
mpre = np.flip(np.maximum.accumulate(np.flip(mpre)))
|
||||
|
||||
# Integrate area under curve
|
||||
method = "interp" # methods: 'continuous', 'interp'
|
||||
if method == "interp":
|
||||
x = np.linspace(0, 1, 101) # 101-point interp (COCO)
|
||||
ap = np.trapz(np.interp(x, mrec, mpre), x) # integrate
|
||||
else: # 'continuous'
|
||||
i = np.where(mrec[1:] != mrec[:-1])[0] # points where x axis (recall) changes
|
||||
ap = np.sum((mrec[i + 1] - mrec[i]) * mpre[i + 1]) # area under curve
|
||||
|
||||
return ap, mpre, mrec
|
||||
Reference in new issue
Block a user