单目3D目标检测——MonoCon 模型训练 | 模型推理

import os
import shutil
# 【一】、读取train.txt文件
with open('./ImageSets/train.txt', 'r') as file:
    # 逐行读取train.txt文件中的文件名ID
    file_ids = [line.strip() for line in file]
# 【1】calib
# 指定路径A和路径B
path_A = './training/calib'
path_B = './train/calib'
# 如果路径B不存在，创建它
if not os.path.exists(path_B):
    os.makedirs(path_B)
# 遍历文件名ID并复制文件到路径B
for file_id in file_ids:
    source_file = os.path.join(path_A, f"{file_id}.txt")
    destination_file = os.path.join(path_B, f"{file_id}.txt")
    if os.path.exists(source_file):
        shutil.copy(source_file, destination_file)
    else:
        print(f"文件未找到：{file_id}.txt")
# 【2】image_2
# 指定路径A和路径B
path_A = './training/image_2'
path_B = './train/image_2'
# 如果路径B不存在，创建它
if not os.path.exists(path_B):
    os.makedirs(path_B)
# 遍历文件名ID并复制文件到路径B
for file_id in file_ids:
    source_file = os.path.join(path_A, f"{file_id}.png")
    destination_file = os.path.join(path_B, f"{file_id}.png")
    if os.path.exists(source_file):
        shutil.copy(source_file, destination_file)
    else:
        print(f"文件未找到：{file_id}.txt")
# 【3】label_2
# 指定路径A和路径B
path_A = './training/label_2'
path_B = './train/label_2'
# 如果路径B不存在，创建它
if not os.path.exists(path_B):
    os.makedirs(path_B)
# 遍历文件名ID并复制文件到路径B
for file_id in file_ids:
    source_file = os.path.join(path_A, f"{file_id}.txt")
    destination_file = os.path.join(path_B, f"{file_id}.txt")
    if os.path.exists(source_file):
        shutil.copy(source_file, destination_file)
    else:
        print(f"文件未找到：{file_id}.txt")
# 【二】、读取valtxt文件
with open('./ImageSets/val.txt', 'r') as file:
    # 逐行读取val.txt文件中的文件名ID
    file_ids = [line.strip() for line in file]
# 【1】calib
# 指定路径A和路径B
path_A = './training/calib'
path_B = './val/calib'
# 如果路径B不存在，创建它
if not os.path.exists(path_B):
    os.makedirs(path_B)
# 遍历文件名ID并复制文件到路径B
for file_id in file_ids:
    source_file = os.path.join(path_A, f"{file_id}.txt")
    destination_file = os.path.join(path_B, f"{file_id}.txt")
    if os.path.exists(source_file):
        shutil.copy(source_file, destination_file)
    else:
        print(f"文件未找到：{file_id}.txt")
# 【2】image_2
# 指定路径A和路径B
path_A = './training/image_2'
path_B = './val/image_2'
# 如果路径B不存在，创建它
if not os.path.exists(path_B):
    os.makedirs(path_B)
# 遍历文件名ID并复制文件到路径B
for file_id in file_ids:
    source_file = os.path.join(path_A, f"{file_id}.png")
    destination_file = os.path.join(path_B, f"{file_id}.png")
    if os.path.exists(source_file):
        shutil.copy(source_file, destination_file)
    else:
        print(f"文件未找到：{file_id}.txt")
# 【3】label_2
# 指定路径A和路径B
path_A = './training/label_2'
path_B = './val/label_2'
# 如果路径B不存在，创建它
if not os.path.exists(path_B):
    os.makedirs(path_B)
# 遍历文件名ID并复制文件到路径B
for file_id in file_ids:
    source_file = os.path.join(path_A, f"{file_id}.txt")
    destination_file = os.path.join(path_B, f"{file_id}.txt")
    if os.path.exists(source_file):
        shutil.copy(source_file, destination_file)
    else:
        print(f"文件未找到：{file_id}.txt")

from yacs.config import CfgNode as CN
_C = CN()
_C.VERSION = 'v1.0.3'
_C.DESCRIPTION = "MonoCon Default Configuration"
_C.OUTPUT_DIR = "./checkpoints_train"                               # Output Directory
_C.SEED = -1                                     # -1: Random Seed Selection
_C.GPU_ID = 0                                    # Index of GPU to use
_C.USE_BENCHMARK = False                          # Value of 'torch.backends.cudnn.benchmark' and 'torch.backends.cudnn.enabled'
# Data
_C.DATA = CN()
_C.DATA.ROOT = r'./dataset'                  # KITTI Root
_C.DATA.BATCH_SIZE = 8
_C.DATA.NUM_WORKERS = 4
_C.DATA.TRAIN_SPLIT = 'train' 
_C.DATA.TEST_SPLIT = 'val' 
_C.DATA.FILTER = CN()
_C.DATA.FILTER.MIN_HEIGHT = 25
_C.DATA.FILTER.MIN_DEPTH = 2
_C.DATA.FILTER.MAX_DEPTH = 65
_C.DATA.FILTER.MAX_TRUNCATION = 0.5
_C.DATA.FILTER.MAX_OCCLUSION = 2
# Model
_C.MODEL = CN()
_C.MODEL.BACKBONE = CN()
_C.MODEL.BACKBONE.NUM_LAYERS = 34
_C.MODEL.BACKBONE.IMAGENET_PRETRAINED = True
_C.MODEL.HEAD = CN()
_C.MODEL.HEAD.NUM_CLASSES = 3
_C.MODEL.HEAD.MAX_OBJS = 30
# Optimization
_C.SOLVER = CN()
_C.SOLVER.OPTIM = CN()
_C.SOLVER.OPTIM.LR = 2.25E-04
_C.SOLVER.OPTIM.WEIGHT_DECAY = 1E-05
_C.SOLVER.OPTIM.NUM_EPOCHS = 20        # Max Training Epochs 200
_C.SOLVER.SCHEDULER = CN()
_C.SOLVER.SCHEDULER.ENABLE = True
_C.SOLVER.CLIP_GRAD = CN()
_C.SOLVER.CLIP_GRAD.ENABLE = True
_C.SOLVER.CLIP_GRAD.NORM_TYPE = 2.0
_C.SOLVER.CLIP_GRAD.MAX_NORM = 35 
# Period
_C.PERIOD = CN()
_C.PERIOD.EVAL_PERIOD = 10                      # In Epochs / Set -1 if you don't want validation 10
_C.PERIOD.LOG_PERIOD = 50                       # In Steps 50

[2023-10-14 15:52:59] Using Random Seed 729445219
==================== Engine Info ====================
- Root: ./checkpoints_train
- Version: v1.0.3
- Description: MonoCon Default Configuration
- Seed: 729445219
- Device: GPU 0 (NVIDIA GeForce GTX 1080 Ti)
- Model: MonoConDetector (# Params: 19620261)
- Optimizer: AdamW
- Scheduler: CyclicScheduler
- Epoch Progress: 1/20
- # Train Samples: 3712
- # Test Samples: 3769
=====================================================
[2023-10-14 15:53:01] Training will be proceeded from epoch 1 to epoch 20.
[2023-10-14 15:53:01] Result files will be saved to './checkpoints_train'.
==================================== Epoch   1 /  20 =====================================
| Progress [##                  ][10.78%] | LR 0.000226 | Loss  25.4260 (721.0474) |
| Progress [####                ][21.55%] | LR 0.000229 | Loss  17.2905 (371.2354) |
| Progress [######              ][32.33%] | LR 0.000233 | Loss  16.1168 ( 19.6029) |
| Progress [########            ][43.10%] | LR 0.000240 | Loss  14.2096 ( 16.5191) |
| Progress [##########          ][53.88%] | LR 0.000248 | Loss  14.2127 ( 14.4836) |
| Progress [############        ][64.66%] | LR 0.000258 | Loss  11.6645 ( 13.4507) |
| Progress [###############     ][75.43%] | LR 0.000269 | Loss  11.5625 ( 13.1914) |
| Progress [#################   ][86.21%] | LR 0.000283 | Loss  12.2610 ( 13.2794) |
| Progress [################### ][96.98%] | LR 0.000298 | Loss  14.5919 ( 13.3250) |
- Average Loss: 91.130
- Epoch Time: 0:08:09
- Remain Time: 2:35:01
- Estimated End-Time: 2023-10-14 18:36:11
==================================== Epoch   2 /  20 =====================================
| Progress [#                   ][ 7.76%] | LR 0.000315 | Loss  10.3478 ( 12.9630) |
| Progress [###                 ][18.53%] | LR 0.000333 | Loss  11.1136 ( 12.7164) |
| Progress [#####               ][29.31%] | LR 0.000353 | Loss  11.2275 ( 12.5311) |
| Progress [########            ][40.09%] | LR 0.000375 | Loss  12.8767 ( 12.3347) |
| Progress [##########          ][50.86%] | LR 0.000398 | Loss  10.9840 ( 12.4795) |
| Progress [############        ][61.64%] | LR 0.000423 | Loss  10.6545 ( 12.4311) |
| Progress [##############      ][72.41%] | LR 0.000449 | Loss  11.1646 ( 12.1970) |
| Progress [################    ][83.19%] | LR 0.000476 | Loss  10.6329 ( 12.2642) |
| Progress [##################  ][93.97%] | LR 0.000505 | Loss  11.2371 ( 12.2401) |
- Average Loss: 12.390
- Epoch Time: 0:08:10
- Remain Time: 2:26:56
- Estimated End-Time: 2023-10-14 18:36:17
==================================== Epoch   3 /  20 =====================================
| Progress [                    ][ 4.74%] | LR 0.000536 | Loss  11.9816 ( 11.9349) |
| Progress [###                 ][15.52%] | LR 0.000567 | Loss  11.9809 ( 11.6493) |
| Progress [#####               ][26.29%] | LR 0.000600 | Loss  12.4710 ( 11.5947) |

python test.py --config_file checkpoints_train/config.yaml --checkpoint_file checkpoints_train/checkpoints/epoch_010.pth --visualize --save_dir save_output --gpu_id 0

python test.py  --config_file pretrained_model/config.yaml --checkpoint_file pretrained_model/best.pth --visualize  --save_dir save_output --gpu_id 0