Skip to content

P
PaddleDetection

s920243400 / PaddleDetection
与 Fork 源项目一致

Fork自 PaddlePaddle / PaddleDetection

通知 2
Star 0
Fork 0
P
PaddleDetection
未验证 提交 9ca73458 编写于 作者: F Feng Ni 提交者： GitHub
浏览文件
操作

[MOT] fix mot readme and cfgs (#4440) 

* remove hardnet85 cfgs, fix readme of mot vehicle pedestrian

* remove hardnet cfg, test=document_fix
上级 2513f35a
隐藏空白更改
内联 并排
Showing with 37 addition and 94 deletion
configs/mot/README.md
浏览文件 @ 9ca73458
......@@ -144,17 +144,14 @@ If you use a stronger detection model, you can get better results. Each txt is t
| backbone | input shape | MOTA | IDF1 | IDS | FP | FN | FPS | download | config |
| :--------------| :------- | :----: | :----: | :----: | :----: | :----: | :------: | :----: |:-----: |
| DLA-34 | 1088x608 | 75.9 | 74.7 | 1021 | 11425 | 31475 | - |[model](https://paddledet.bj.bcebos.com/models/mot/fairmot_enhance_dla34_30e_1088x608.pdparams) | [config](./fairmot_enhance_dla34_30e_1088x608.yml) |
| HarDNet-85 | 1088x608 | 75.0 | 70.0 | 1050 | 11837 | 32774 | - |[model](https://paddledet.bj.bcebos.com/models/mot/fairmot_enhance_hardnet85_30e_1088x608.pdparams) | [config](./fairmot/fairmot_enhance_hardnet85_30e_1088x608.yml) |
### Results on MOT-17 Test Set
| backbone | input shape | MOTA | IDF1 | IDS | FP | FN | FPS | download | config |
| :--------------| :------- | :----: | :----: | :----: | :----: | :----: | :------: | :----: |:-----: |
| DLA-34 | 1088x608 | 75.3 | 74.2 | 3270 | 29112 | 106749 | - |[model](https://paddledet.bj.bcebos.com/models/mot/fairmot_enhance_dla34_30e_1088x608.pdparams) | [config](./fairmot_enhance_dla34_30e_1088x608.yml) |
| HarDNet-85 | 1088x608 | 74.7 | 70.7 | 3210 | 29790 | 109914 | - |[model](https://paddledet.bj.bcebos.com/models/mot/fairmot_enhance_hardnet85_30e_1088x608.pdparams) | [config](./fairmot/fairmot_enhance_hardnet85_30e_1088x608.yml) |
**Notes:**
FairMOT enhance DLA-34 used 8 GPUs for training and mini-batch size as 16 on each GPU，and trained for 60 epoches. The crowdhuman dataset is added to the train-set during training.
FairMOT enhance HarDNet-85 used 8 GPUs for training and mini-batch size as 10 on each GPU，and trained for 30 epoches. The crowdhuman dataset is added to the train-set during training.
### FairMOT light model
......
configs/mot/README_cn.md
浏览文件 @ 9ca73458
......@@ -143,17 +143,14 @@ wget https://dataset.bj.bcebos.com/mot/det_results_dir.zip
| 骨干网络 | 输入尺寸 | MOTA | IDF1 | IDS | FP | FN | FPS | 下载链接 | 配置文件 |
| :--------------| :------- | :----: | :----: | :----: | :----: | :----: | :------: | :----: |:-----: |
| DLA-34 | 1088x608 | 75.9 | 74.7 | 1021 | 11425 | 31475 | - |[下载链接](https://paddledet.bj.bcebos.com/models/mot/fairmot_enhance_dla34_60e_1088x608.pdparams) | [配置文件](./fairmot_enhance_dla34_60e_1088x608.yml) |
| HarDNet-85 | 1088x608 | 75.0 | 70.0 | 1050 | 11837 | 32774 | - |[下载链接](https://paddledet.bj.bcebos.com/models/mot/fairmot_enhance_hardnet85_30e_1088x608.pdparams) | [配置文件](./fairmot/fairmot_enhance_hardnet85_30e_1088x608.yml) |
### 在MOT-17 Test Set上结果
| 骨干网络 | 输入尺寸 | MOTA | IDF1 | IDS | FP | FN | FPS | 下载链接 | 配置文件 |
| :--------------| :------- | :----: | :----: | :----: | :----: | :----: | :------: | :----: |:-----: |
| DLA-34 | 1088x608 | 75.3 | 74.2 | 3270 | 29112 | 106749 | - |[下载链接](https://paddledet.bj.bcebos.com/models/mot/fairmot_enhance_dla34_60e_1088x608.pdparams) | [配置文件](./fairmot_enhance_dla34_60e_1088x608.yml) |
| HarDNet-85 | 1088x608 | 74.7 | 70.7 | 3210 | 29790 | 109914 | - |[下载链接](https://paddledet.bj.bcebos.com/models/mot/fairmot_enhance_hardnet85_30e_1088x608.pdparams) | [配置文件](./fairmot/fairmot_enhance_hardnet85_30e_1088x608.yml) |
**注意:**
FairMOT enhance DLA-34使用8个GPU进行训练，每个GPU上batch size为16，训练60个epoch，并且训练集中加入了crowdhuman数据集一起参与训练。
FairMOT enhance HarDNet-85 使用8个GPU进行训练，每个GPU上batch size为10，训练30个epoch，并且训练集中加入了crowdhuman数据集一起参与训练。
### FairMOT轻量级模型
......
configs/mot/fairmot/README.md
浏览文件 @ 9ca73458
......@@ -42,17 +42,14 @@ English | [简体中文](README_cn.md)
| backbone | input shape | MOTA | IDF1 | IDS | FP | FN | FPS | download | config |
| :--------------| :------- | :----: | :----: | :----: | :----: | :----: | :------: | :----: |:-----: |
| DLA-34 | 1088x608 | 75.9 | 74.7 | 1021 | 11425 | 31475 | - |[model](https://paddledet.bj.bcebos.com/models/mot/fairmot_enhance_dla34_60e_1088x608.pdparams) | [config](./fairmot_enhance_dla34_60e_1088x608.yml) |
| HarDNet-85 | 1088x608 | 75.0 | 70.0 | 1050 | 11837 | 32774 | - |[model](https://paddledet.bj.bcebos.com/models/mot/fairmot_enhance_hardnet85_30e_1088x608.pdparams) | [config](./fairmot_enhance_hardnet85_30e_1088x608.yml) |
### Results on MOT-17 Test Set
| backbone | input shape | MOTA | IDF1 | IDS | FP | FN | FPS | download | config |
| :--------------| :------- | :----: | :----: | :----: | :----: | :----: | :------: | :----: |:-----: |
| DLA-34 | 1088x608 | 75.3 | 74.2 | 3270 | 29112 | 106749 | - |[model](https://paddledet.bj.bcebos.com/models/mot/fairmot_enhance_dla34_60e_1088x608.pdparams) | [config](./fairmot_enhance_dla34_60e_1088x608.yml) |
| HarDNet-85 | 1088x608 | 74.7 | 70.7 | 3210 | 29790 | 109914 | - |[model](https://paddledet.bj.bcebos.com/models/mot/fairmot_enhance_hardnet85_30e_1088x608.pdparams) | [config](./fairmot_enhance_hardnet85_30e_1088x608.yml) |
**Notes:**
FairMOT enhance DLA-34 used 8 GPUs for training and mini-batch size as 16 on each GPU，and trained for 60 epoches. The crowdhuman dataset is added to the train-set during training.
FairMOT enhance HarDNet-85 used 8 GPUs for training and mini-batch size as 10 on each GPU，and trained for 30 epoches. The crowdhuman dataset is added to the train-set during training.
### FairMOT light model
......
configs/mot/fairmot/README_cn.md
浏览文件 @ 9ca73458
......@@ -41,17 +41,14 @@
| 骨干网络 | 输入尺寸 | MOTA | IDF1 | IDS | FP | FN | FPS | 下载链接 | 配置文件 |
| :--------------| :------- | :----: | :----: | :----: | :----: | :----: | :------: | :----: |:-----: |
| DLA-34 | 1088x608 | 75.9 | 74.7 | 1021 | 11425 | 31475 | - |[下载链接](https://paddledet.bj.bcebos.com/models/mot/fairmot_enhance_dla34_60e_1088x608.pdparams) | [配置文件](./fairmot_enhance_dla34_60e_1088x608.yml) |
| HarDNet-85 | 1088x608 | 75.0 | 70.0 | 1050 | 11837 | 32774 | - |[下载链接](https://paddledet.bj.bcebos.com/models/mot/fairmot_enhance_hardnet85_30e_1088x608.pdparams) | [配置文件](./fairmot_enhance_hardnet85_30e_1088x608.yml) |
### 在MOT-17 Test Set上结果
| 骨干网络 | 输入尺寸 | MOTA | IDF1 | IDS | FP | FN | FPS | 下载链接 | 配置文件 |
| :--------------| :------- | :----: | :----: | :----: | :----: | :----: | :------: | :----: |:-----: |
| DLA-34 | 1088x608 | 75.3 | 74.2 | 3270 | 29112 | 106749 | - |[下载链接](https://paddledet.bj.bcebos.com/models/mot/fairmot_enhance_dla34_60e_1088x608.pdparams) | [配置文件](./fairmot_enhance_dla34_60e_1088x608.yml) |
| HarDNet-85 | 1088x608 | 74.7 | 70.7 | 3210 | 29790 | 109914 | - |[下载链接](https://paddledet.bj.bcebos.com/models/mot/fairmot_enhance_hardnet85_30e_1088x608.pdparams) | [配置文件](./fairmot_enhance_hardnet85_30e_1088x608.yml) |
**注意:**
FairMOT enhance DLA-34使用8个GPU进行训练，每个GPU上batch size为16，训练60个epoch，并且训练集中加入了crowdhuman数据集一起参与训练。
FairMOT enhance HarDNet-85 使用8个GPU进行训练，每个GPU上batch size为10，训练30个epoch，并且训练集中加入了crowdhuman数据集一起参与训练。
### FairMOT轻量级模型
......
configs/mot/fairmot/fairmot_enhance_hardnet85_30e_1088x608.yml 已删除 100644 → 0
浏览文件 @ 2513f35a
_BASE_: [
'../../datasets/mot.yml',
'../../runtime.yml',
'_base_/optimizer_30e.yml',
'_base_/fairmot_hardnet85.yml',
'_base_/fairmot_reader_1088x608.yml',
]
norm_type: sync_bn
use_ema: true
ema_decay: 0.9998
worker_num: 4
TrainReader:
inputs_def:
image_shape: [3, 608, 1088]
sample_transforms:
- Decode: {}
- RGBReverse: {}
- AugmentHSV: {}
- LetterBoxResize: {target_size: [608, 1088]}
- MOTRandomAffine: {reject_outside: False}
- RandomFlip: {}
- BboxXYXY2XYWH: {}
- NormalizeBox: {}
- NormalizeImage: {mean: [0, 0, 0], std: [1, 1, 1]}
- RGBReverse: {}
- Permute: {}
batch_transforms:
- Gt2FairMOTTarget: {}
batch_size: 10
shuffle: True
drop_last: True
use_shared_memory: True
epoch: 30
LearningRate:
base_lr: 0.0001
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [20,]
use_warmup: False
OptimizerBuilder:
optimizer:
type: Adam
regularizer: NULL
weights: output/fairmot_enhance_hardnet85_30e_1088x608/model_final
configs/mot/pedestrian/README_cn.md
浏览文件 @ 9ca73458
......@@ -20,7 +20,7 @@
| VisDrone | 1088x608 | 49.2 | 63.1 | - | [下载链接](https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608_visdrone_pedestrian.pdparams) | [配置文件](./fairmot_dla34_30e_1088x608_visdrone_pedestrian.yml) |
**注意:**
FairMOT均使用DLA-34为骨干网络，2个GPU进行训练，每个GPU上batch size为6，训练30个epoch。
FairMOT均使用DLA-34为骨干网络，4个GPU进行训练，每个GPU上batch size为6，训练30个epoch。
## 数据集准备和处理
......@@ -48,8 +48,10 @@ tools/visdrone/visdrone2mot.py: 生成visdrone_pedestrian据集
│ ├── visdrone2mot.py
│ ├── visdrone_pedestrian
│ │ ├── images
│ │ │ ├── train
│ │ │ ├── val
│ │ ├── labels_with_ids
│ │ │ ├── train
│ │ │ ├── val
# 执行
python visdrone2mot.py --transMot=True --data_name=visdrone_pedestrian --phase=val
......@@ -61,36 +63,36 @@ python visdrone2mot.py --transMot=True --data_name=visdrone_pedestrian --phase=t
### 1. 训练
使用2个GPU通过如下命令一键式启动训练
```bash
python -m paddle.distributed.launch --log_dir=./fairmot_dla34_30e_1088x608_pathtrack/ --gpus 0,1 tools/train.py -c configs/mot/pedestrian/fairmot_dla34_30e_1088x608_pathtrack.yml
python -m paddle.distributed.launch --log_dir=./fairmot_dla34_30e_1088x608_visdrone_pedestrian/ --gpus 0,1 tools/train.py -c configs/mot/pedestrian/fairmot_dla34_30e_1088x608_visdrone_pedestrian.yml
```
### 2. 评估
使用单张GPU通过如下命令一键式启动评估
```bash
# 使用PaddleDetection发布的权重
CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/pedestrian/fairmot_dla34_30e_1088x608_pathtrack.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608_pathtrack.pdparams
CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/pedestrian/fairmot_dla34_30e_1088x608_visdrone_pedestrian.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608_visdrone_pedestrian.pdparams
# 使用训练保存的checkpoint
CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/pedestrian/fairmot_dla34_30e_1088x608_pathtrack.yml -o weights=output/fairmot_dla34_30e_1088x608_pathtrack/model_final.pdparams
CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/pedestrian/fairmot_dla34_30e_1088x608_visdrone_pedestrian.yml -o weights=output/fairmot_dla34_30e_1088x608_visdrone_pedestrian/model_final.pdparams
```
### 3. 预测
使用单个GPU通过如下命令预测一个视频，并保存为视频
```bash
# 预测一个视频
CUDA_VISIBLE_DEVICES=0 python tools/infer_mot.py -c configs/mot/pedestrian/fairmot_dla34_30e_1088x608_pathtrack.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608_pathtrack.pdparams --video_file={your video name}.mp4 --save_videos
CUDA_VISIBLE_DEVICES=0 python tools/infer_mot.py -c configs/mot/pedestrian/fairmot_dla34_30e_1088x608_visdrone_pedestrian.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608_visdrone_pedestrian.pdparams --video_file={your video name}.mp4 --save_videos
```
**注意:**
请先确保已经安装了[ffmpeg](https://ffmpeg.org/ffmpeg.html), Linux(Ubuntu)平台可以直接用以下命令安装：`apt-get update && apt-get install -y ffmpeg`。
### 4. 导出预测模型
```bash
CUDA_VISIBLE_DEVICES=0 python tools/export_model.py -c configs/mot/pedestrian/fairmot_dla34_30e_1088x608_pathtrack.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608_pathtrack.pdparams
CUDA_VISIBLE_DEVICES=0 python tools/export_model.py -c configs/mot/pedestrian/fairmot_dla34_30e_1088x608_visdrone_pedestrian.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608_visdrone_pedestrian.pdparams
```
### 5. 用导出的模型基于Python去预测
```bash
python deploy/python/mot_jde_infer.py --model_dir=output_inference/fairmot_dla34_30e_1088x608_pathtrack --video_file={your video name}.mp4 --device=GPU --save_mot_txts
python deploy/python/mot_jde_infer.py --model_dir=output_inference/fairmot_dla34_30e_1088x608_visdrone_pedestrian --video_file={your video name}.mp4 --device=GPU --save_mot_txts
```
**注意:**
跟踪模型是对视频进行预测，不支持单张图的预测，默认保存跟踪结果可视化后的视频，可添加`--save_mot_txts`表示保存跟踪结果的txt文件，或`--save_images`表示保存跟踪结果可视化图片。
......
configs/mot/vehicle/README_cn.md
浏览文件 @ 9ca73458
......@@ -40,16 +40,16 @@ tools/bdd100kmot/gen_labels_MOT.py：生成单类别的labels_with_ids文件
tools/visdrone/visdrone2mot.py：生成visdrone_vehicle
```
### 2、bdd100k_vehicle数据集处理
### 2、bdd100kmot_vehicle数据集处理
```
# 复制tools/bdd100kmot里的代码到数据集目录下
# 生成bdd100kmot_vehicle MOT格式的数据，抽取类别classes=2,3,4,9,10 (car, truck, bus, trailer, other vehicle)
<<--生成前目录-->>
├── bdd100k_path
├── bdd100k
│ ├── images
│ ├── labels
<<--生成后目录-->>
├── bdd100k_path
├── bdd100k
│ ├── images
│ ├── labels
│ ├── bdd100kmot_vehicle
......@@ -77,10 +77,12 @@ sh gen_bdd100kmot_vehicle.sh
│ ├── annotations
│ ├── sequences
│ ├── visdrone2mot.py
│ ├── visdrone-vehicle
│ ├── visdrone_vehicle
│ │ ├── images
│ │ │ ├── train
│ │ │ ├── val
│ │ ├── labels_with_ids
│ │ │ ├── train
│ │ │ ├── val
# 执行
python visdrone2mot.py --transMot=True --data_name=visdrone_vehicle --phase=val
......@@ -92,36 +94,36 @@ python visdrone2mot.py --transMot=True --data_name=visdrone_vehicle --phase=trai
### 1. 训练
使用2个GPU通过如下命令一键式启动训练
```bash
python -m paddle.distributed.launch --log_dir=./fairmot_dla34_30e_1088x608_bdd100k_vehicle/ --gpus 0,1 tools/train.py -c configs/mot/vehicle/fairmot_dla34_30e_1088x608_bdd100k_vehicle.yml
python -m paddle.distributed.launch --log_dir=./fairmot_dla34_30e_1088x608_bdd100kmot_vehicle/ --gpus 0,1 tools/train.py -c configs/mot/vehicle/fairmot_dla34_30e_1088x608_bdd100kmot_vehicle.yml
```
### 2. 评估
使用单张GPU通过如下命令一键式启动评估
```bash
# 使用PaddleDetection发布的权重
CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/vehicle/fairmot_dla34_30e_1088x608_bdd100k_vehicle.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608_bdd100k_vehicle.pdparams
CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/vehicle/fairmot_dla34_30e_1088x608_bdd100kmot_vehicle.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608_bdd100kmot_vehicle.pdparams
# 使用训练保存的checkpoint
CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/vehicle/fairmot_dla34_30e_1088x608_bdd100k_vehicle.yml -o weights=output/fairmot_dla34_30e_1088x608_bdd100k_vehicle/model_final.pdparams
CUDA_VISIBLE_DEVICES=0 python tools/eval_mot.py -c configs/mot/vehicle/fairmot_dla34_30e_1088x608_bdd100kmot_vehicle.yml -o weights=output/fairmot_dla34_30e_1088x608_bdd100kmot_vehicle/model_final.pdparams
```
### 3. 预测
使用单个GPU通过如下命令预测一个视频，并保存为视频
```bash
# 预测一个视频
CUDA_VISIBLE_DEVICES=0 python tools/infer_mot.py -c configs/mot/vehicle/fairmot_dla34_30e_1088x608_bdd100k_vehicle.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608_bdd100k_vehicle.pdparams --video_file={your video name}.mp4 --save_videos
CUDA_VISIBLE_DEVICES=0 python tools/infer_mot.py -c configs/mot/vehicle/fairmot_dla34_30e_1088x608_bdd100kmot_vehicle.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608_bdd100kmot_vehicle.pdparams --video_file={your video name}.mp4 --save_videos
```
**注意:**
请先确保已经安装了[ffmpeg](https://ffmpeg.org/ffmpeg.html), Linux(Ubuntu)平台可以直接用以下命令安装：`apt-get update && apt-get install -y ffmpeg`。
### 4. 导出预测模型
```bash
CUDA_VISIBLE_DEVICES=0 python tools/export_model.py -c configs/mot/vehicle/fairmot_dla34_30e_1088x608_bdd100k_vehicle.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608_bdd100k_vehicle.pdparams
CUDA_VISIBLE_DEVICES=0 python tools/export_model.py -c configs/mot/vehicle/fairmot_dla34_30e_1088x608_bdd100kmot_vehicle.yml -o weights=https://paddledet.bj.bcebos.com/models/mot/fairmot_dla34_30e_1088x608_bdd100kmot_vehicle.pdparams
```
### 5. 用导出的模型基于Python去预测
```bash
python deploy/python/mot_jde_infer.py --model_dir=output_inference/fairmot_dla34_30e_1088x608_bdd100k_vehicle --video_file={your video name}.mp4 --device=GPU --save_mot_txts
python deploy/python/mot_jde_infer.py --model_dir=output_inference/fairmot_dla34_30e_1088x608_bdd100kmot_vehicle --video_file={your video name}.mp4 --device=GPU --save_mot_txts
```
**注意:**
跟踪模型是对视频进行预测，不支持单张图的预测，默认保存跟踪结果可视化后的视频，可添加`--save_mot_txts`表示保存跟踪结果的txt文件，或`--save_images`表示保存跟踪结果可视化图片。
......
configs/mot/vehicle/tools/bdd100kmot/bdd100k2mot.py
浏览文件 @ 9ca73458
......@@ -19,6 +19,8 @@ import cv2
import random
import numpy as np
import argparse
import tqdm
import json
def mkdir_if_missing(d):
......@@ -62,7 +64,7 @@ def bdd2mot_tracking(img_dir, label_dir, save_img_dir, save_label_dir):
os.system('cp {} {}'.format(source_img, target_img))
def transBBOx(bbox):
def transBbox(bbox):
# bbox --> cx cy w h
bbox = list(map(lambda x: float(x), bbox))
bbox[0] = (bbox[0] - bbox[2] / 2) * 1280
......@@ -106,10 +108,9 @@ def genSingleImageMot(inputPath, classes=[]):
mot_line = []
mot_line.append(id_line[-1])
mot_line.append(str(id_idx))
id_line_temp = transBBOx(id_line[2:6])
id_line_temp = transBbox(id_line[2:6])
mot_line.extend(id_line_temp)
mot_line.append('1')
# mot_line.append(id_line[0]) # origin class
mot_line.append('1') # origin class: id_line[0]
mot_line.append('1') # permanent class => 1
mot_line.append('1')
mot_gt.append(mot_line)
......@@ -211,7 +212,7 @@ def VisualGt(dataPath, phase='train'):
img_list_path = sorted(glob.glob(osp.join(seqPath, 'img1', '*.jpg')))
imgIndex = random.randint(0, len(img_list_path))
img_Path = img_list_path[imgIndex]
#
frame_value = img_Path.split('/')[-1].replace('.jpg', '')
frame_value = frame_value.split('-')[-1]
frame_value = int(frame_value)
......@@ -290,18 +291,17 @@ def copyImg(fromRootPath, toRootPath, phase):
if __name__ == "__main__":
parser = argparse.ArgumentParser(description='BDD100K to MOT format')
parser.add_argument("--data_path", default='/path/to/bdd100k')
parser.add_argument("--data_path", default='bdd100k')
parser.add_argument("--phase", default='train')
parser.add_argument("--classes", default='2,3,4,9,10')
parser.add_argument("--img_dir", default="/path/to/bdd/image/")
parser.add_argument("--label_dir", default="/path/to/bdd/label/")
parser.add_argument("--save_path", default="/save/path")
parser.add_argument("--img_dir", default="bdd100k/images/track/")
parser.add_argument("--label_dir", default="bdd100k/labels/box_track_20/")
parser.add_argument("--save_path", default="bdd100kmot_vehicle")
parser.add_argument("--height", default=720)
parser.add_argument("--width", default=1280)
args = parser.parse_args()
### for bdd tracking dataset
attr_dict = dict()
attr_dict["categories"] = [{
"supercategory": "none",
......@@ -350,7 +350,7 @@ if __name__ == "__main__":
}]
attr_id_dict = {i['name']: i['id'] for i in attr_dict['categories']}
# create BDD training set tracking in MOT format
# create bdd100kmot_vehicle training set in MOT format
print('Loading and converting training set...')
train_img_dir = os.path.join(args.img_dir, 'train')
train_label_dir = os.path.join(args.label_dir, 'train')
......@@ -361,7 +361,7 @@ if __name__ == "__main__":
bdd2mot_tracking(train_img_dir, train_label_dir, save_img_dir,
save_label_dir)
# create BDD validation set tracking in MOT format
# create bdd100kmot_vehicle validation set in MOT format
print('Loading and converting validation set...')
val_img_dir = os.path.join(args.img_dir, 'val')
val_label_dir = os.path.join(args.label_dir, 'val')
......
configs/mot/vehicle/tools/bdd100kmot/gen_bdd100kmot_vehicle.sh
浏览文件 @ 9ca73458
data_path=bdd100k_path
img_dir=${data_path}/images
label_dir=${data_path}/labels
data_path=bdd100k
img_dir=${data_path}/images/track
label_dir=${data_path}/labels/box_track_20
save_path=${data_path}/bdd100kmot_vehicle
phasetrain=train
......
configs/mot/vehicle/tools/bdd100kmot/gen_labels_MOT.py
浏览文件 @ 9ca73458
......@@ -26,7 +26,7 @@ def mkdirs(d):
if __name__ == "__main__":
parser = argparse.ArgumentParser(description='BDD100K to MOT format')
parser.add_argument(
"--mot_data", default='/paddle/dataset/bdd100kmot/bdd100k_small')
"--mot_data", default='./bdd100k')
parser.add_argument("--phase", default='train')
args = parser.parse_args()
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或