# OpenPCDet-Interpretability

**Repository Path**: discover304/open-pcdet-interpretability

## Basic Information

- **Project Name**: OpenPCDet-Interpretability
- **Description**: Tsinghua
- **Primary Language**: Python
- **License**: Apache-2.0
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2023-07-04
- **Last Updated**: 2023-07-16

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# OpenPCDet-Interpretability

> Detailed Instruction: [README_official.md](README_official.md).

## Contributing

Instruction about collaboration.

---
Coding:

1. Create new branch.
2. Add, commit and push to your branch.
3. Create pull request, merging into main branch in case you believe your code is finished.

---

Bugs:

1. Create nuw issue.
2. Report in group chat (wechat or slack)

---

## Commands

Commands used to start testing or training process.

---
Prepare datasets

```bash
USER_NAME="" # TODO your user name

cd open-pcdet-interpretability
mkdir -p data/nuscenes/v1.0-trainval
ln -s /home/tsing-adept/datasets/nuScenes/* /home/${USER_NAME}/Env/open-pcdet-interpretability/data/nuscenes/v1.0-trainval
```
---

Test:

```bash
CONFIG_FILE="tools/cfgs/nuscenes_models/bevfusion.yaml"
BATCH_SIZE=32 # TODO your user name
CKPT="" # TODO path to checkpoint
NUM_GPUS=4  # TODO number of gpu

# TODO change gpus not occupied by others
CUDA_VISIBLE_DEVICES=0, 1, 3, 4, 7\
sh scripts/dist_test.sh \
    ${NUM_GPUS} \
    --cfg_file ${CONFIG_FILE} \
    --batch_size ${BATCH_SIZE} \
    --ckpt ${CKPT}
```
---

Train + Eval
```bash
CONFIG_FILE="tools/cfgs/nuscenes_models/bevfusion.yaml"
BATCH_SIZE=32
CKPT="" # FIXME parameter not list in official doc
NUM_GPUS=4

CUDA_VISIBLE_DEVICES=1, 6, 7, 8 \
sh scripts/dist_train.sh \
    ${NUM_GPUS} \
    --cfg_file ${CONFIG_FILE} \
    --batch_size ${BATCH_SIZE} \
    --ckpt ${CKPT} # FIXME parameter not list in official doc
```
---