yancie-yjr / DBQ-SSD

The official implementation of the paper DBQ-SSD: Dynamic Ball Query for Efficient 3D Object Detection (ICLR 2023)
Apache License 2.0
18 stars 3 forks source link

kitti performance #7

Open ljwwwiop opened 8 months ago

ljwwwiop commented 8 months ago

I'm following the code and config file you provided on the KITTI data and it still doesn't seem to achieve the performance on the validation set you mentioned in your paper, can you provide me with your checkpoint or are there any other tuning parameters? And I found that the dynamic group configuration in the code repository is not used now, is it also not used in your best checkpoint. I hope you can answer this!

ljwwwiop commented 8 months ago
        DYNAMIC: [True, True, True, False, False, True]
        #DYNAMIC_COST: [1.262796015, 0.94903174, 1.130387545, 0, 0, 1.922637284]
        DYNAMIC_COST: [[0.276546776, 1.117151827], [0.342942774, 0.693246126], [0.324052423, 0.869811475], [0, 0], [0, 0], [0.391717553, 1.609101295]]
        DYNAMIC_GROUP: [False, False, False, False, False, False]

this is default config yaml file

ljwwwiop commented 8 months ago

Car AP@0.70, 0.70, 0.70: bbox AP:93.5697, 89.8340, 89.1830 bev AP:89.9567, 87.8019, 85.6441 3d AP:88.5625, 78.7357, 77.4495 aos AP:93.53, 89.74, 88.97 Car AP_R40@0.70, 0.70, 0.70: bbox AP:96.8774, 94.0185, 91.7220 bev AP:93.4930, 88.4263, 87.3645 3d AP:89.9583, 81.3058, 77.1416 aos AP:96.84, 93.89, 91.48 Car AP@0.70, 0.50, 0.50: bbox AP:93.5697, 89.8340, 89.1830 bev AP:93.6142, 93.1825, 89.5333 3d AP:93.5658, 89.9640, 89.4682 aos AP:93.53, 89.74, 88.97 Car AP_R40@0.70, 0.50, 0.50: bbox AP:96.8774, 94.0185, 91.7220 bev AP:96.9845, 96.0161, 94.1601 3d AP:96.9405, 94.5340, 93.9973 aos AP:96.84, 93.89, 91.48 Pedestrian AP@0.50, 0.50, 0.50: bbox AP:70.2216, 68.3792, 63.3181 bev AP:62.8441, 59.3165, 53.5329 3d AP:59.7327, 54.7102, 50.3246 aos AP:65.00, 62.74, 57.99 Pedestrian AP_R40@0.50, 0.50, 0.50: bbox AP:70.8842, 68.1660, 63.5042 bev AP:63.2960, 58.6851, 53.0203 3d AP:59.2694, 54.2282, 48.5055 aos AP:65.16, 62.13, 57.55 Pedestrian AP@0.50, 0.25, 0.25: bbox AP:70.2216, 68.3792, 63.3181 bev AP:76.7348, 74.3060, 70.2011 3d AP:76.7316, 74.2777, 70.0905 aos AP:65.00, 62.74, 57.99 Pedestrian AP_R40@0.50, 0.25, 0.25: bbox AP:70.8842, 68.1660, 63.5042 bev AP:78.7909, 76.1684, 70.7676 3d AP:78.7848, 76.1269, 70.4055 aos AP:65.16, 62.13, 57.55 Cyclist AP@0.50, 0.50, 0.50: bbox AP:89.8611, 78.6869, 75.9146 bev AP:87.8073, 73.2507, 69.0345 3d AP:85.8559, 70.1521, 67.4281 aos AP:89.70, 78.28, 75.39 Cyclist AP_R40@0.50, 0.50, 0.50: bbox AP:95.3015, 80.5684, 77.3563 bev AP:92.8199, 74.1536, 69.9502 3d AP:90.2432, 70.9028, 67.1232 aos AP:95.08, 80.11, 76.77 Cyclist AP@0.50, 0.25, 0.25: bbox AP:89.8611, 78.6869, 75.9146 bev AP:94.6091, 77.3793, 73.4238 3d AP:94.6091, 77.3768, 73.4238 aos AP:89.70, 78.28, 75.39 Cyclist AP_R40@0.50, 0.25, 0.25: bbox AP:95.3015, 80.5684, 77.3563 bev AP:96.3169, 78.5624, 74.6746 3d AP:96.3169, 78.5061, 74.6335 aos AP:95.08, 80.11, 76.77

results

ljwwwiop commented 8 months ago

image