def parse_args():
parser = argparse.ArgumentParser(
"Submitit script", parents=[classification.get_args_parser()])
parser.add_argument("--ngpus",
default=8,
type=int,
help="Number of gpus to request on each node")
parser.add_argument("--nodes",
default=1,
type=int,
help="Number of nodes to request")
parser.add_argument("--timeout",
default=2800,
type=int,
help="Duration of the job")
parser.add_argument("--job_dir",
default="",
type=str,
help="Job dir. Leave empty for automatic.")
parser.add_argument("--partition",
default="learnlab",
type=str,
help="Partition where to submit")
parser.add_argument("--use_volta32",
action='store_true',
help="Big models? Use this")
parser.add_argument(
'--comment',
default="",
type=str,
help='Comment to pass to scheduler, e.g. priority message')
return parser.parse_args()
def parse_args():
detection_parser = detection.get_args_parser()
parser = argparse.ArgumentParser("Submitit for detection",
parents=[detection_parser])
parser.add_argument("--ngpus",
default=2,
type=int,
help="Number of gpus to request on each node")
parser.add_argument("--nodes",
default=2,
type=int,
help="Number of nodes to request")
parser.add_argument("--timeout",
default=3000,
type=int,
help="Duration of the job")
parser.add_argument("--job_dir",
default="",
type=str,
help="Job dir. Leave empty for automatic.")
parser.add_argument(
"--mail",
default="",
type=str,
help="Email this user when the job finishes if specified")
return parser.parse_args()
def get_args_parser():
detection_parser = detection.get_args_parser()
parser = argparse.ArgumentParser(
"Get predictions for GQA and dump to file",
parents=[detection_parser],
add_help=False)
parser.add_argument("--split",
type=str,
default="testdev",
choices=("testdev", "test", "challenge", "submission"))
return parser
def local_get_args_parser():
detection_parser = detection.get_args_parser()
parser = argparse.ArgumentParser(
"Get predictions for clevr and dump to file",
parents=[detection_parser],
add_help=False)
parser.add_argument("--split",
type=str,
default="val",
choices=("val", "test", "testA", "testB", "valA",
"valB"))
parser.add_argument("--clevr_eval_path", type=str, default="")
return parser
def parse_args():
detection_parser = detection.get_args_parser()
parser = argparse.ArgumentParser('Submitit for detection', parents=[
detection_parser])
parser.add_argument('--ngpus', default=8, type=int, help=\
'Number of gpus to request on each node')
parser.add_argument('--nodes', default=4, type=int, help=\
'Number of nodes to request')
parser.add_argument('--timeout', default=60, type=int, help=\
'Duration of the job')
parser.add_argument('--job_dir', default='', type=str, help=\
'Job dir. Leave empty for automatic.')
return parser.parse_args()
def parse_args():
detection_parser = detection.get_args_parser()
parser = argparse.ArgumentParser(
"Submitit for detection", parents=[detection_parser]
)
parser.add_argument(
"--ngpus", default=4, type=int, help="Number of gpus to request on each node"
)
parser.add_argument(
"--nodes", default=1, type=int, help="Number of nodes to request"
)
parser.add_argument(
"--hours", default=12, type=int, help="Duration of the job in hours"
)
parser.add_argument(
"--job_dir", default="", type=str, help="Job dir. Leave empty for automatic."
)
parser.add_argument("--experiment_name", type=str, default="test_experiment_delete")
parser.add_argument("--constraint", type=str, default=None)
return parser.parse_args()
from PIL import Image
import requests
import matplotlib.pyplot as plt
import cv2
import torch
from torch import nn
from torchvision.models import resnet50
from models import build
import torchvision.transforms as T
from main import get_args_parser
import numpy as np
torch.set_grad_enabled(False)
parser = get_args_parser()
# for output bounding box post-processing
def box_cxcywh_to_xyxy(x):
x_c, y_c, w, h = x.unbind(1)
b = [(x_c - 0.5 * w), (y_c - 0.5 * h), (x_c + 0.5 * w), (y_c + 0.5 * h)]
return torch.stack(b, dim=1)
def rescale_bboxes(out_bbox, size):
img_w, img_h = size
b = box_cxcywh_to_xyxy(out_bbox)
b = b * torch.tensor([img_w, img_h, img_w, img_h], dtype=torch.float32)
return b
self.write_results(txt_path=os.path.join(self.predict_path,
f'{self.seq_num}.txt'),
frame_id=(i + 1),
bbox_xyxy=tracker_outputs[:, :4],
identities=tracker_outputs[:, 5])
filter_pub_det(
os.path.join(self.predict_path, f'{self.seq_num}.txt'),
f'/data/Dataset/mot/MOT17/images/test/{self.seq_num}/det/det.txt')
print("totally {} dts {} occlusion dts".format(total_dts,
total_occlusion_dts))
if __name__ == '__main__':
parser = argparse.ArgumentParser('DETR training and evaluation script',
parents=[get_args_parser()])
args = parser.parse_args()
if args.output_dir:
Path(args.output_dir).mkdir(parents=True, exist_ok=True)
# load model and weights
detr, _, _ = build_model(args)
checkpoint = torch.load(args.resume, map_location='cpu')
detr = load_model(detr, args.resume)
detr.eval()
detr = detr.cuda()
# '''for MOT17 submit'''
sub_dir = 'MOT17/images/test'
seq_nums = [
'MOT17-01-SDP', 'MOT17-03-SDP', 'MOT17-06-SDP', 'MOT17-07-SDP',
# for visual
# self.visualize_img_with_bbox(cur_vis_img_path, ori_img, dt_instances, ref_pts=all_ref_pts, gt_boxes=gt_boxes)
tracker_outputs = self.tr_tracker.update(dt_instances)
self.write_results(txt_path=os.path.join(self.predict_path, f'{self.seq_num}.txt'),
frame_id=(i + 1),
bbox_xyxy=tracker_outputs[:, :4],
identities=tracker_outputs[:, 5])
filter_pub_det(os.path.join(self.predict_path, f'{self.seq_num}.txt'),
f'/data/Dataset/mot/MOT17/images/test/{self.seq_num}/det/det.txt')
print("totally {} dts {} occlusion dts".format(total_dts, total_occlusion_dts))
if __name__ == '__main__':
parser = argparse.ArgumentParser('DETR training and evaluation script', parents=[get_args_parser()])
args = parser.parse_args()
if args.output_dir:
Path(args.output_dir).mkdir(parents=True, exist_ok=True)
# load model and weights
detr, _, _ = build_model(args)
checkpoint = torch.load(args.resume, map_location='cpu')
detr = load_model(detr, args.resume)
detr.eval()
detr = detr.cuda()
# '''for MOT17 submit'''
sub_dir = 'MOT17/images/test'
seq_nums = ['MOT17-01-SDP',
'MOT17-03-SDP',
def get_args_parser():
detection_parser = detection.get_args_parser()
parser = argparse.ArgumentParser("Evaluate MDETR on LVIS detection", parents=[detection_parser], add_help=False)
parser.add_argument("--lvis_minival_path", type=str, default="")
return parser
