Bienqq
/
car-detection-bayes
2
1
Fork 0
Code Issues Pull Requests Releases Wiki Activity

mAP recorded during training

This commit is contained in:
Glenn Jocher 2018-11-14 15:14:41 +00:00
parent 9dbc3ec1c4
commit 45c5567723
4 changed files with 107 additions and 96 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

180
test.py
View File

@ -11,7 +11,7 @@ parser.add_argument('-data_config_path', type=str, default='cfg/coco.data', help
parser.add_argument('-weights_path', type=str, default='weights/yolov3.pt', help='path to weights file') parser.add_argument('-weights_path', type=str, default='weights/yolov3.pt', help='path to weights file')
parser.add_argument('-class_path', type=str, default='data/coco.names', help='path to class label file') parser.add_argument('-class_path', type=str, default='data/coco.names', help='path to class label file')
parser.add_argument('-iou_thres', type=float, default=0.5, help='iou threshold required to qualify as detected') parser.add_argument('-iou_thres', type=float, default=0.5, help='iou threshold required to qualify as detected')
parser.add_argument('-conf_thres', type=float, default=0.5, help='object confidence threshold') parser.add_argument('-conf_thres', type=float, default=0.3, help='object confidence threshold')
parser.add_argument('-nms_thres', type=float, default=0.45, help='iou threshold for non-maximum suppression') parser.add_argument('-nms_thres', type=float, default=0.45, help='iou threshold for non-maximum suppression')
parser.add_argument('-n_cpu', type=int, default=0, help='number of cpu threads to use during batch generation') parser.add_argument('-n_cpu', type=int, default=0, help='number of cpu threads to use during batch generation')
parser.add_argument('-img_size', type=int, default=416, help='size of each image dimension') parser.add_argument('-img_size', type=int, default=416, help='size of each image dimension')
@ -21,112 +21,118 @@ print(opt)
cuda = torch.cuda.is_available() cuda = torch.cuda.is_available()
device = torch.device('cuda:0' if cuda else 'cpu') device = torch.device('cuda:0' if cuda else 'cpu')
# Configure run
data_config = parse_data_config(opt.data_config_path)
num_classes = int(data_config['classes'])
if platform == 'darwin': # MacOS (local)
test_path = data_config['valid']
else: # linux (cloud, i.e. gcp)
test_path = '../coco/5k.part'
# Initiate model def main(opt):
model = Darknet(opt.cfg, opt.img_size) # Configure run
data_config = parse_data_config(opt.data_config_path)
nC = int(data_config['classes']) # number of classes (80 for COCO)
if platform == 'darwin': # MacOS (local)
test_path = data_config['valid']
else: # linux (cloud, i.e. gcp)
test_path = '../coco/5k.part'
# Load weights # Initiate model
if opt.weights_path.endswith('.weights'): # darknet format model = Darknet(opt.cfg, opt.img_size)
load_weights(model, opt.weights_path)
elif opt.weights_path.endswith('.pt'): # pytorch format
checkpoint = torch.load(opt.weights_path, map_location='cpu')
model.load_state_dict(checkpoint['model'])
del checkpoint
model.to(device).eval() # Load weights
if opt.weights_path.endswith('.weights'): # darknet format
load_weights(model, opt.weights_path)
elif opt.weights_path.endswith('.pt'): # pytorch format
checkpoint = torch.load(opt.weights_path, map_location='cpu')
model.load_state_dict(checkpoint['model'])
del checkpoint
# Get dataloader model.to(device).eval()
# dataset = load_images_with_labels(test_path)
# dataloader = torch.utils.data.DataLoader(dataset, batch_size=opt.batch_size, shuffle=False, num_workers=opt.n_cpu)
dataloader = load_images_and_labels(test_path, batch_size=opt.batch_size, img_size=opt.img_size)
print('Compute mAP...') # Get dataloader
# dataset = load_images_with_labels(test_path)
# dataloader = torch.utils.data.DataLoader(dataset, batch_size=opt.batch_size, shuffle=False, num_workers=opt.n_cpu)
dataloader = load_images_and_labels(test_path, batch_size=opt.batch_size, img_size=opt.img_size)
nC = 80 # number of classes print('Compute mAP...')
correct = 0
targets = None
outputs, mAPs, TP, confidence, pred_class, target_class = [], [], [], [], [], []
AP_accum, AP_accum_count = np.zeros(nC), np.zeros(nC)
for batch_i, (imgs, targets) in enumerate(dataloader):
imgs = imgs.to(device)
with torch.no_grad(): mAP = 0
output = model(imgs) outputs, mAPs, TP, confidence, pred_class, target_class = [], [], [], [], [], []
output = non_max_suppression(output, conf_thres=opt.conf_thres, nms_thres=opt.nms_thres) AP_accum, AP_accum_count = np.zeros(nC), np.zeros(nC)
for batch_i, (imgs, targets) in enumerate(dataloader):
imgs = imgs.to(device)
# Compute average precision for each sample with torch.no_grad():
for sample_i in range(len(targets)): output = model(imgs)
correct = [] output = non_max_suppression(output, conf_thres=opt.conf_thres, nms_thres=opt.nms_thres)
# Get labels for sample where width is not zero (dummies) # Compute average precision for each sample
annotations = targets[sample_i] for sample_i in range(len(targets)):
# Extract detections correct = []
detections = output[sample_i]
if detections is None: # Get labels for sample where width is not zero (dummies)
# If there are no detections but there are annotations mask as zero AP annotations = targets[sample_i]
if annotations.size(0) != 0: # Extract detections
detections = output[sample_i]
if detections is None:
# If there are no detections but there are annotations mask as zero AP
if annotations.size(0) != 0:
mAPs.append(0)
continue
# Get detections sorted by decreasing confidence scores
detections = detections[np.argsort(-detections[:, 4])]
# If no annotations add number of detections as incorrect
if annotations.size(0) == 0:
# correct.extend([0 for _ in range(len(detections))])
mAPs.append(0) mAPs.append(0)
continue continue
else:
target_cls = annotations[:, 0]
# Get detections sorted by decreasing confidence scores # Extract target boxes as (x1, y1, x2, y2)
detections = detections[np.argsort(-detections[:, 4])] target_boxes = xywh2xyxy(annotations[:, 1:5])
target_boxes *= opt.img_size
# If no annotations add number of detections as incorrect detected = []
if annotations.size(0) == 0: for *pred_bbox, conf, obj_conf, obj_pred in detections:
target_cls = []
# correct.extend([0 for _ in range(len(detections))])
mAPs.append(0)
continue
else:
target_cls = annotations[:, 0]
# Extract target boxes as (x1, y1, x2, y2) pred_bbox = torch.FloatTensor(pred_bbox).view(1, -1)
target_boxes = xywh2xyxy(annotations[:, 1:5]) # Compute iou with target boxes
target_boxes *= opt.img_size iou = bbox_iou(pred_bbox, target_boxes)
# Extract index of largest overlap
best_i = np.argmax(iou)
# If overlap exceeds threshold and classification is correct mark as correct
if iou[best_i] > opt.iou_thres and obj_pred == annotations[best_i, 0] and best_i not in detected:
correct.append(1)
detected.append(best_i)
else:
correct.append(0)
detected = [] # Compute Average Precision (AP) per class
for *pred_bbox, conf, obj_conf, obj_pred in detections: AP, AP_class = ap_per_class(tp=correct, conf=detections[:, 4], pred_cls=detections[:, 6],
target_cls=target_cls)
pred_bbox = torch.FloatTensor(pred_bbox).view(1, -1) # Accumulate AP per class
# Compute iou with target boxes AP_accum_count += np.bincount(AP_class, minlength=nC)
iou = bbox_iou(pred_bbox, target_boxes) AP_accum += np.bincount(AP_class, minlength=nC, weights=AP)
# Extract index of largest overlap
best_i = np.argmax(iou)
# If overlap exceeds threshold and classification is correct mark as correct
if iou[best_i] > opt.iou_thres and obj_pred == annotations[best_i, 0] and best_i not in detected:
correct.append(1)
detected.append(best_i)
else:
correct.append(0)
# Compute Average Precision (AP) per class # Compute mean AP for this image
AP, AP_class = ap_per_class(tp=correct, conf=detections[:, 4], pred_cls=detections[:, 6], target_cls=target_cls) mAP = AP.mean()
# Accumulate AP per class # Append image mAP to list
AP_accum_count += np.bincount(AP_class, minlength=nC) mAPs.append(mAP)
AP_accum += np.bincount(AP_class, minlength=nC, weights=AP)
# Compute mean AP for this image # Print image mAP and running mean mAP
mAP = AP.mean() print(
'+ Sample [%d/%d] AP: %.4f (%.4f)' % (len(mAPs), len(dataloader) * opt.batch_size, mAP, np.mean(mAPs)))
# Append image mAP to list # Print mAP per class
mAPs.append(mAP) classes = load_classes(opt.class_path) # Extracts class labels from file
for i, c in enumerate(classes):
print('%15s: %-.4f' % (c, AP_accum[i] / AP_accum_count[i]))
# Print image mAP and running mean mAP # Print mAP
print('+ Sample [%d/%d] AP: %.4f (%.4f)' % (len(mAPs), len(dataloader) * opt.batch_size, mAP, np.mean(mAPs))) print('Mean Average Precision: %.4f' % np.mean(mAPs))
return mAP
# Print mAP per class
classes = load_classes(opt.class_path) # Extracts class labels from file
for i, c in enumerate(classes):
print('%15s: %-.4f' % (c, AP_accum[i] / AP_accum_count[i]))
# Print mAP if __name__ == '__main__':
print('Mean Average Precision: %.4f' % np.mean(mAPs)) mAP = main(opt)

19
train.py
View File

@ -1,5 +1,6 @@
import argparse import argparse
import time import time
import test
from models import * from models import *
from utils.datasets import * from utils.datasets import *
@ -103,10 +104,10 @@ def main(opt):
# scheduler.step() # scheduler.step()
# Update scheduler (manual) at 0, 54, 61 epochs to 1e-3, 1e-4, 1e-5 # Update scheduler (manual) at 0, 54, 61 epochs to 1e-3, 1e-4, 1e-5
if epoch < 50: if epoch > 50:
lr = 1e-4
else:
lr = 1e-5 lr = 1e-5
else:
lr = 1e-4
for g in optimizer.param_groups: for g in optimizer.param_groups:
g['lr'] = lr g['lr'] = lr
@ -160,10 +161,6 @@ def main(opt):
t1 = time.time() t1 = time.time()
print(s) print(s)
# Write epoch results
with open('results.txt', 'a') as file:
file.write(s + '\n')
# Update best loss # Update best loss
loss_per_target = rloss['loss'] / rloss['nT'] loss_per_target = rloss['loss'] / rloss['nT']
if loss_per_target < best_loss: if loss_per_target < best_loss:
@ -184,6 +181,14 @@ def main(opt):
if (epoch > 0) & (epoch % 5 == 0): if (epoch > 0) & (epoch % 5 == 0):
os.system('cp weights/latest.pt weights/backup' + str(epoch) + '.pt') os.system('cp weights/latest.pt weights/backup' + str(epoch) + '.pt')
# Calculate mAP
test.opt.weights_path = 'weights/latest.pt'
mAP = test.main(test.opt)
# Write epoch results
with open('results.txt', 'a') as file:
file.write(s + '%11.3g' % mAP + '\n')
# Save final model # Save final model
dt = time.time() - t0 dt = time.time() - t0
print('Finished %g epochs in %.2fs (%.2fs/epoch)' % (epoch, dt, dt / (epoch + 1))) print('Finished %g epochs in %.2fs (%.2fs/epoch)' % (epoch, dt, dt / (epoch + 1)))

2
utils/gcp.sh
View File

@ -11,7 +11,7 @@ gsutil cp gs://ultralytics/yolov3.pt yolov3/weights
python3 detect.py python3 detect.py
# Test # Test
python3 test.py -img_size 416 -weights_path weights/latest.pt -conf_thres 0.1 python3 test.py -img_size 416 -weights_path weights/latest.pt -conf_thres 0.5
# Download and Test # Download and Test
sudo rm -rf yolov3 && git clone https://github.com/ultralytics/yolov3 && cd yolov3 sudo rm -rf yolov3 && git clone https://github.com/ultralytics/yolov3 && cd yolov3

2
utils/utils.py
View File

@ -435,7 +435,7 @@ def plot_results():
import numpy as np import numpy as np
import matplotlib.pyplot as plt import matplotlib.pyplot as plt
plt.figure(figsize=(16, 8)) plt.figure(figsize=(16, 8))
s = ['X', 'Y', 'Width', 'Height', 'Objectness', 'Classification', 'Total Loss', 'Precision', 'Recall'] s = ['X', 'Y', 'Width', 'Height', 'Objectness', 'Classification', 'Total Loss', 'Precision', 'Recall', 'mAP']
for f in ('results.txt',): for f in ('results.txt',):
results = np.loadtxt(f, usecols=[2, 3, 4, 5, 6, 7, 8, 9, 10]).T results = np.loadtxt(f, usecols=[2, 3, 4, 5, 6, 7, 8, 9, 10]).T
for i in range(9): for i in range(9):