tensorrt检测加跟踪:
YoloV5_JDE_TensorRT_for_Track:基于张量的多目标跟踪库-源码_jde多目标跟踪-其它代码类资源-CSDN下载
关于tensorboard:
安装:
torch1.8.0以上,自带tensorboard
低版本可能没有,单独安装:
pip install tensorboardX
调用:
from tensorboardXimport SummaryWriter
这个参数感觉不需要,去掉了。
self.training |= self.export
https://github.com/ultralytics/yolov5
测试结果:
1060上,640*640 模型27m yolov5s batch size 12是ok的
1060上,640*640 模型85m yolov5m batch size 8是ok的
测试代码:
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--cfg', type=str, default='yolov5s.yaml', help='model.yaml')
parser.add_argument('--device', default='0', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')
opt = parser.parse_args()
opt.cfg = glob.glob('./**/' + opt.cfg, recursive=True)[0] # find file
device = torch_utils.select_device(opt.device)
# Create model
model = Model(opt.cfg).to(device)
model.eval()
torch.save(model.state_dict(), f'v2.pth')
input = torch.randn(12, 3, 640, 640).cuda()
for i in range(10):
start = time.time()
output = model(input)
print('output.size ', time.time() - start, output[0].size())
del output