File size: 1,504 Bytes
6a5bb7e | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 | """深度分析mAP50-95瓶颈：per-IoU threshold + per-size breakdown"""
import sys, os
os.chdir("/home/user/goat")
sys.path.insert(0, "/home/user/goat")
import numpy as np
print("Importing YOLO...", flush=True)
from ultralytics import YOLO
print("Loading model...", flush=True)
model = YOLO('runs/detect/Detection_experiments/v6_1_s_refined/weights/best.pt')
print("Running val...", flush=True)
results = model.val(data='Data/Detection_dataset/dataset.yaml', imgsz=1536, batch=2, verbose=False)
print("mAP50: {:.4f}".format(results.box.map50), flush=True)
print("mAP50-95: {:.4f}".format(results.box.map), flush=True)
# Check available attributes
attrs = [x for x in dir(results.box) if not x.startswith('_')]
print("Available attrs: {}".format(attrs), flush=True)
# Try per-threshold AP
try:
ap = results.box.ap
print("ap shape: {}".format(ap.shape), flush=True)
print("ap values: {}".format(ap), flush=True)
except Exception as e:
print("ap error: {}".format(e), flush=True)
try:
all_ap = results.box.all_ap
print("all_ap shape: {}".format(all_ap.shape), flush=True)
# all_ap is typically [num_classes, num_iou_thresholds]
iou_thresholds = np.arange(0.5, 1.0, 0.05)
for i, t in enumerate(iou_thresholds):
if i < all_ap.shape[-1]:
print(" IoU={:.2f}: AP={:.4f}".format(t, all_ap[0, i] if all_ap.ndim > 1 else all_ap[i]), flush=True)
except Exception as e:
print("all_ap error: {}".format(e), flush=True)
print("\nDONE", flush=True)
|