Hello, I want to test the model on the nusence data set. How should I write the configuration file?

[libingDY]

Hello, I want to test the model on the nusence data set. How should I write the configuration file?

[NiranjanRavi1993]

@libingDY Have you had success? I was able to train nuscenes but evaluation throws annotation errors.

[libingDY]

Yes I get it working but my accuracy is very low

[NiranjanRavi1993]

I am seeing the opposite. For some categories, error value is greater than 1 in nuscenes dataset.

[libingDY]

Can you send me a copy of your configuration file? What was the result of the nusence data set you finally ran?

[NiranjanRavi1993]

Sure, not sure if its right way. But I just utilized class names and ANCHOR_GENERATOR_CONFIG from here: https://github.com/open-mmlab/OpenPCDet/blob/master/tools/cfgs/nuscenes_models/cbgs_second_multihead.yaml

Is that how you also proceeded?

[libingDY]

My configuration is as follows：

CLASS_NAMES: ['car']

DATA_CONFIG:
BASE_CONFIG: cfgs/dataset_configs/nuscenes_dataset.yaml

MODEL:
NAME: RDIoUNet

VFE:
    NAME: MeanVFE

BACKBONE_3D:
    NAME: VoxelBackBone8xLightweight

MAP_TO_BEV:
    NAME: HeightCompression
    NUM_BEV_FEATURES: 256

BACKBONE_2D:
    NAME: newCTBEVBackbone

    LAYER_NUMS: [ 1, 0 ]
    LAYER_STRIDES: [ 2, 2 ]
    NUM_FILTERS: [ 128, 256 ]
    UPSAMPLE_STRIDES: [ 1, 2 ]
    NUM_UPSAMPLE_FILTERS: [ 256, 256 ]

DENSE_HEAD:
    NAME: AnchorHeadRDIoU
    CLASS_AGNOSTIC: False

    USE_DIRECTION_CLASSIFIER: True
    DIR_OFFSET: 0.78539
    DIR_LIMIT_OFFSET: 0.0
    NUM_DIR_BINS: 2

    ANCHOR_GENERATOR_CONFIG: [
        {
            'class_name': car,
            'anchor_sizes': [[4.63, 1.97, 1.74]],
            'anchor_rotations': [0, 1.57],
            'anchor_bottom_heights': [-0.95],
            'align_center': False,
            'feature_map_stride': 8,
            'matched_threshold': 0.6,
            'unmatched_threshold': 0.45
        },
    ]

    TARGET_ASSIGNER_CONFIG:
        NAME: AxisAlignedTargetAssigner
        POS_FRACTION: -1.0
        SAMPLE_SIZE: 512
        NORM_BY_NUM_EXAMPLES: False
        MATCH_HEIGHT: False
        BOX_CODER: ResidualCoder



    LOSS_CONFIG:
        LOSS_WEIGHTS: {
            'cls_weight': 1.0,
            'loc_weight': 2.0,
            'dir_weight': 0.2,
            'code_theta_weights': [1.0],
        }

POST_PROCESSING:
    RECALL_THRESH_LIST: [0.3, 0.5, 0.7]
    SCORE_THRESH: 0.4
    OUTPUT_RAW_SCORE: False

    EVAL_METRIC: kitti

    NMS_CONFIG:
        MULTI_CLASSES_NMS: False
        NMS_TYPE: nms_gpu
        NMS_THRESH: 0.2
        NMS_PRE_MAXSIZE: 4096
        NMS_POST_MAXSIZE: 500

OPTIMIZATION:
BATCH_SIZE_PER_GPU: 4
NUM_EPOCHS: 100

OPTIMIZER: adam_onecycle
LR: 0.00035
WEIGHT_DECAY: 0.01
MOMENTUM: 0.9

MOMS: [0.95, 0.85]
PCT_START: 0.4
DIV_FACTOR: 10
DECAY_STEP_LIST: [35, 45]
LR_DECAY: 0.1
LR_CLIP: 0.0000001

LR_WARMUP: False
WARMUP_EPOCH: 1

GRAD_NORM_CLIP: 10

[libingDY]

my email is libing12014014@163.com，We can quickly communicate this issue via the email。