Skip to content

Latest commit

 

History

History
153 lines (88 loc) · 3.99 KB

get_started.md

File metadata and controls

153 lines (88 loc) · 3.99 KB
Raw

Get Started

This page will lead you to go through our data and code.

Prerequisites

Please follow the steps below to get prepared.

1. Environment

Step 1. Create conda environment and activate it.

conda create --name hdmap python=3.8
conda activate hdmap

Step 2. Install PyTorch.

pip install torch==1.9.0+cu111 torchvision==0.10.0+cu111 -f https://download.pytorch.org/whl/torch_stable.html

Step 3. Install MMCV series.

# Install mmcv-series
pip install mmcv-full==1.3.9 -f https://download.openmmlab.com/mmcv/dist/cu111/torch1.9.0/index.html
pip install mmdet==2.14.0
pip install mmsegmentation==0.14.1

Step 4. Install mmdetection3d.

Currently we are usingmmdetection3d of version 0.17.3 . To install mmdet3d, please first download the releases of 0.17.3 from https://github.com/open-mmlab/mmdetection3d/releases. Then run

cd mmdetection3d-0.17.3
pip install -v -e .

For more details about installation, please refer to open-mmlab getting_started.md.

Step 5. Install other requirements.

pip install -r requirements.txt

2. Data

Please see data.md.

Baseline Model

We provide a baseline model based on VectorMapNet (Checkpoint).

Expected Results

Method mAP Ped Crossing Divider Boundary Config File Ckpt
VectorMapNet 42.09 38.87 49.23 38.15 config

Baseline Training

Single GPU training

python tools/train.py src/configs/vectormapnet.py

Multi GPU training

bash tools/dist_train.sh src/configs/vectormapnet.py ${NUM_GPUS}

Evaluation

For details of evaluation metrics, please see metrics.md.

Evaluating a Checkpoint

Single GPU evaluation

python tools/test.py src/configs/vectormapnet.py ${CHECKPOINT} --split val --eval

Generate a submission file on test set without evaluation:

python tools/test.py src/configs/vectormapnet.py ${CHECKPOINT} --split test --format-only

Multi GPU evaluation

bash tools/dist_test.sh src/configs/vectormapnet.py ${CHECKPOINT} ${NUM_GPUS} --split val --eval

Tips for options:

  1. Use --split option to specify the dataset split to test.
  2. Use --eval option to generate a submission file and run evaluation code. Only available when you are testing on validation set (with --split val set).
  3. Use --format-only option to generate a submission file without evaluation. By setting --split test --format-only you will get a submission file which can be directly submitted to our test server.

Evaluating a submission file on validation set

If you want evaluate a submission file (only available on validation set since you do not have annotations on test set), run

python tools/evaluate_submission.py ${SUBMISSION_FILE} ${VAL_ANNOTATION_FILE}

Visualization

Visualization tools can be used to visualize ground-truth labels and model's prediction results. Map polylines will be projected (with z-axis set to 0 if missing) and rendered on surrounding camera images. A 2D BEV map view will also be rendered.

To visualize ground-truth labels of a log specified by ${LOG_ID} 

python tools/visualize.py ${LOG_ID} ${ANN_FILE}

for example,

python tools/visualization/visualize.py 10001 data/val_annotations.json

To visualize also prediction results and filter out all predictions with score lower than ${THR}

python tools/visualize.py ${LOG_ID} ${ANN_FILE} --result ${SUBMISSION_FILE} --thr ${THR}

You can specify where to save these rendered images by setting --out-dir, default to demo/.