File size: 6,666 Bytes
67bb36a |
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 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 |
import argparse
import os, sys
import shutil
import time
from pathlib import Path
import imageio
BASE_DIR = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
sys.path.append(BASE_DIR)
print(sys.path)
import cv2
import torch
import torch.backends.cudnn as cudnn
from numpy import random
import scipy.special
import numpy as np
import torchvision.transforms as transforms
import PIL.Image as image
from lib.config import cfg
from lib.config import update_config
from lib.utils.utils import create_logger, select_device, time_synchronized
from lib.models import get_net
from lib.dataset import LoadImages, LoadStreams
from lib.core.general import non_max_suppression, scale_coords
from lib.utils import plot_one_box,show_seg_result
from lib.core.function import AverageMeter
from lib.core.postprocess import morphological_process
from tqdm import tqdm
normalize = transforms.Normalize(
mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]
)
transform=transforms.Compose([
transforms.ToTensor(),
normalize,
])
def detect(cfg,opt):
logger, final_output_dir, tb_log_dir = create_logger(
cfg, cfg.LOG_DIR, 'demo')
device = select_device(logger,opt.device)
if os.path.exists(opt.save_dir): # output dir
shutil.rmtree(opt.save_dir) # delete dir
os.makedirs(opt.save_dir) # make new dir
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = get_net(cfg)
checkpoint = torch.load(opt.weights, map_location= device)
model.load_state_dict(checkpoint['state_dict'])
model = model.to(device)
if half:
model.half() # to FP16
# Set Dataloader
if opt.source.isnumeric():
cudnn.benchmark = True # set True to speed up constant image size inference
dataset = LoadStreams(opt.source, img_size=opt.img_size)
bs = len(dataset) # batch_size
else:
dataset = LoadImages(opt.source, img_size=opt.img_size)
bs = 1 # batch_size
# Get names and colors
names = model.module.names if hasattr(model, 'module') else model.names
colors = [[random.randint(0, 255) for _ in range(3)] for _ in range(len(names))]
# Run inference
t0 = time.time()
vid_path, vid_writer = None, None
img = torch.zeros((1, 3, opt.img_size, opt.img_size), device=device) # init img
_ = model(img.half() if half else img) if device.type != 'cpu' else None # run once
model.eval()
inf_time = AverageMeter()
nms_time = AverageMeter()
for i, (path, img, img_det, vid_cap,shapes) in tqdm(enumerate(dataset),total = len(dataset)):
img = transform(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
det_out, da_seg_out,ll_seg_out= model(img)
t2 = time_synchronized()
inf_out,train_out = det_out
inf_time.update(t2-t1,img.size(0))
# Apply NMS
t3 = time_synchronized()
det_pred = non_max_suppression(inf_out, conf_thres=0.25, iou_thres=0.45, classes=None, agnostic=False)
t4 = time_synchronized()
nms_time.update(t4-t3,img.size(0))
det=det_pred[0]
save_path = str(opt.save_dir +'/'+ Path(path).name) if dataset.mode != 'stream' else str(opt.save_dir + '/' + "web.mp4")
_, _, height, width = img.shape
h,w,_=img_det.shape
pad_w, pad_h = shapes[1][1]
pad_w = int(pad_w)
pad_h = int(pad_h)
ratio = shapes[1][0][1]
da_predict = da_seg_out[:, :, pad_h:(height-pad_h),pad_w:(width-pad_w)]
da_seg_mask = torch.nn.functional.interpolate(da_predict, scale_factor=int(1/ratio), mode='bilinear')
_, da_seg_mask = torch.max(da_seg_mask, 1)
da_seg_mask = da_seg_mask.int().squeeze().cpu().numpy()
ll_predict = ll_seg_out[:, :,pad_h:(height-pad_h),pad_w:(width-pad_w)]
ll_seg_mask = torch.nn.functional.interpolate(ll_predict, scale_factor=int(1/ratio), mode='bilinear')
_, ll_seg_mask = torch.max(ll_seg_mask, 1)
ll_seg_mask = ll_seg_mask.int().squeeze().cpu().numpy()
img_det = show_seg_result(img_det, (da_seg_mask, ll_seg_mask), _, _, is_demo=True)
if len(det):
det[:,:4] = scale_coords(img.shape[2:],det[:,:4],img_det.shape).round()
for *xyxy,conf,cls in reversed(det):
label_det_pred = f'{names[int(cls)]} {conf:.2f}'
plot_one_box(xyxy, img_det , label=label_det_pred, color=colors[int(cls)], line_thickness=2)
if dataset.mode == 'images':
cv2.imwrite(save_path,img_det)
elif dataset.mode == 'video':
if vid_path != save_path: # new video
vid_path = save_path
if isinstance(vid_writer, cv2.VideoWriter):
vid_writer.release() # release previous video writer
fourcc = 'mp4v' # output video codec
fps = vid_cap.get(cv2.CAP_PROP_FPS)
h,w,_=img_det.shape
vid_writer = cv2.VideoWriter(save_path, cv2.VideoWriter_fourcc(*fourcc), fps, (w, h))
vid_writer.write(img_det)
else:
cv2.imshow('image', img_det)
cv2.waitKey(1) # 1 millisecond
print('Results saved to %s' % Path(opt.save_dir))
print('Done. (%.3fs)' % (time.time() - t0))
print('inf : (%.4fs/frame) nms : (%.4fs/frame)' % (inf_time.avg,nms_time.avg))
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--weights', nargs='+', type=str, default='weights/End-to-end.pth', help='model.pth path(s)')
parser.add_argument('--source', type=str, default='inference/images', help='source') # file/folder ex:inference/images
parser.add_argument('--img-size', type=int, default=640, help='inference size (pixels)')
parser.add_argument('--conf-thres', type=float, default=0.25, help='object confidence threshold')
parser.add_argument('--iou-thres', type=float, default=0.45, help='IOU threshold for NMS')
parser.add_argument('--device', default='cpu', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')
parser.add_argument('--save-dir', type=str, default='inference/output', help='directory to save results')
parser.add_argument('--augment', action='store_true', help='augmented inference')
parser.add_argument('--update', action='store_true', help='update all models')
opt = parser.parse_args()
with torch.no_grad():
detect(cfg,opt)
|