hellcatzm/mask_rcnn_keypoints

将Mask_RCNN应用于关键点检测，基于TensorFlow和Keras

mask_rcnn_keypoints's Issues

你好，我想问一下，如果训练集中的一张图片有两个instance，那数据输入应该怎么处理呢

您好，我已使用1个服装类训练好了模型，运行demo_detect.ipynb脚本查看训练效果时，detections却显示N>1个实例结果，请问会是哪里出现了问题呢？

[[[4.25662994e-02 1.70000000e+02 7.13350117e-01 1.70000000e+02
1.00000000e+00 9.91867006e-01]
[3.30203772e-03 1.70000000e+02 8.16060245e-01 1.70000000e+02
1.00000000e+00 9.89467800e-01]
[6.64100647e-02 1.70000000e+02 7.56213665e-01 1.70000000e+02
1.00000000e+00 9.85076249e-01]
[5.11958003e-02 1.70000000e+02 7.85606384e-01 1.70000000e+02
1.00000000e+00 9.81133044e-01]
[7.27798343e-02 1.70000000e+02 6.37804091e-01 1.70000000e+02
1.00000000e+00 9.79126990e-01]
[0.00000000e+00 1.70000000e+02 8.19039464e-01 1.70000000e+02
1.00000000e+00 9.79049325e-01]
[8.36743712e-02 1.70000000e+02 5.78436792e-01 1.70000000e+02
1.00000000e+00 9.76869524e-01]
[1.07032627e-01 1.70000000e+02 7.77483582e-01 1.70000000e+02
1.00000000e+00 9.76482093e-01]
[8.90105665e-02 1.70000000e+02 6.23850226e-01 1.70000000e+02
1.00000000e+00 9.74576175e-01]
[8.79395753e-02 1.70000000e+02 5.68945587e-01 1.70000000e+02
1.00000000e+00 9.72842753e-01]
[0.00000000e+00 1.70000000e+02 8.64177525e-01 1.70000000e+02
1.00000000e+00 9.72585380e-01]
[7.85277486e-02 1.70000000e+02 8.41955245e-01 1.70000000e+02
1.00000000e+00 9.69959676e-01]
[7.26488829e-02 1.70000000e+02 8.09916496e-01 1.70000000e+02
1.00000000e+00 9.69448745e-01]
[0.00000000e+00 1.70000000e+02 8.00597191e-01 1.70000000e+02
1.00000000e+00 9.69397783e-01]
[3.17474902e-02 1.70000000e+02 8.46101522e-01 1.70000000e+02
1.00000000e+00 9.68068659e-01]
[1.44336596e-01 1.70000000e+02 6.35651827e-01 1.70000000e+02
1.00000000e+00 9.59369481e-01]
[3.35816741e-02 1.70000000e+02 8.04694831e-01 1.70000000e+02
1.00000000e+00 9.57619965e-01]
[2.49698594e-01 1.70000000e+02 7.36236155e-01 1.70000000e+02
1.00000000e+00 9.55162108e-01]
[4.72135544e-01 1.70000000e+02 8.20890546e-01 1.70000000e+02
1.00000000e+00 9.54922557e-01]
[4.40273285e-02 1.70000000e+02 8.79183471e-01 1.70000000e+02
1.00000000e+00 9.54196095e-01]
[4.63822782e-02 1.70000000e+02 9.39643025e-01 1.70000000e+02
1.00000000e+00 9.51709569e-01]
[1.14914775e-02 1.70000000e+02 8.69611263e-01 1.70000000e+02
1.00000000e+00 9.49731588e-01]
[4.82452512e-01 1.70000000e+02 8.55745673e-01 1.70000000e+02
1.00000000e+00 9.45989370e-01]
[6.68278635e-02 1.70000000e+02 6.57619357e-01 1.70000000e+02
1.00000000e+00 9.45289314e-01]
[1.11367613e-01 1.70000000e+02 4.96867269e-01 1.70000000e+02
1.00000000e+00 9.44907606e-01]
[2.69133478e-01 1.70000000e+02 8.16037655e-01 1.70000000e+02
1.00000000e+00 9.44354534e-01]
[4.10076737e-01 1.70000000e+02 8.21716905e-01 1.70000000e+02
1.00000000e+00 9.42657351e-01]
[5.95972240e-02 1.70000000e+02 6.76495671e-01 1.70000000e+02
1.00000000e+00 9.41745162e-01]
[1.19627297e-01 1.70000000e+02 5.37452161e-01 1.70000000e+02
1.00000000e+00 9.38550413e-01]
[9.11914408e-02 1.70000000e+02 6.37079954e-01 1.70000000e+02
1.00000000e+00 9.38170433e-01]
[1.52568549e-01 1.70000000e+02 7.61175036e-01 1.70000000e+02
1.00000000e+00 9.35723960e-01]
[4.45630699e-01 1.70000000e+02 8.97680402e-01 1.70000000e+02
1.00000000e+00 9.34601486e-01]
[0.00000000e+00 1.70000000e+02 8.75895143e-01 1.70000000e+02
1.00000000e+00 9.32018459e-01]
[9.78352427e-02 1.70000000e+02 7.29681790e-01 1.70000000e+02
1.00000000e+00 9.31587577e-01]
[1.14306614e-01 1.70000000e+02 5.54970741e-01 1.70000000e+02
1.00000000e+00 9.30742204e-01]
[1.00092903e-01 1.70000000e+02 5.57554841e-01 1.70000000e+02
1.00000000e+00 9.29118335e-01]
[1.03261262e-01 1.70000000e+02 6.09454632e-01 1.70000000e+02
1.00000000e+00 9.28648472e-01]
[3.74812961e-01 1.70000000e+02 7.90434361e-01 1.70000000e+02
1.00000000e+00 9.26839769e-01]
[1.71488047e-01 1.70000000e+02 6.94176078e-01 1.70000000e+02
1.00000000e+00 9.26217556e-01]
[1.96113139e-01 1.70000000e+02 9.00408983e-01 1.70000000e+02
1.00000000e+00 9.24274683e-01]
[2.58485079e-01 1.70000000e+02 7.18724728e-01 1.70000000e+02
1.00000000e+00 9.21923757e-01]
[7.52372444e-02 1.70000000e+02 6.64920688e-01 1.70000000e+02
1.00000000e+00 9.16667759e-01]
[5.32034874e-01 1.70000000e+02 8.68950009e-01 1.70000000e+02
1.00000000e+00 9.13202107e-01]
[3.69970024e-01 1.70000000e+02 7.49254823e-01 1.70000000e+02
1.00000000e+00 9.11786079e-01]
[1.10356957e-01 1.70000000e+02 5.75300694e-01 1.70000000e+02
1.00000000e+00 9.10579622e-01]
[3.99774253e-01 1.70000000e+02 8.29287410e-01 1.70000000e+02
1.00000000e+00 9.10549760e-01]
[2.75878191e-01 1.70000000e+02 8.95645261e-01 1.70000000e+02
1.00000000e+00 9.08962071e-01]
[1.27184987e-01 1.70000000e+02 4.93220925e-01 1.70000000e+02
1.00000000e+00 9.07884955e-01]
[3.14731240e-01 1.70000000e+02 7.48176217e-01 1.70000000e+02
1.00000000e+00 9.04735506e-01]
[3.29482138e-01 1.70000000e+02 8.09100747e-01 1.70000000e+02
1.00000000e+00 9.01359916e-01]
[0.00000000e+00 1.70000000e+02 7.70210505e-01 1.70000000e+02
1.00000000e+00 9.01145816e-01]
[1.95117891e-01 1.70000000e+02 7.96142459e-01 1.70000000e+02
1.00000000e+00 9.00941253e-01]
[1.94110274e-02 1.70000000e+02 9.01583374e-01 1.70000000e+02
1.00000000e+00 9.00754452e-01]
[1.28993839e-01 1.70000000e+02 8.62857342e-01 1.70000000e+02
1.00000000e+00 9.00615275e-01]
[3.80188525e-02 1.70000000e+02 7.43819356e-01 1.70000000e+02
1.00000000e+00 8.99960935e-01]
[1.40734851e-01 1.70000000e+02 7.95415461e-01 1.70000000e+02
1.00000000e+00 8.96768153e-01]
[4.17237580e-01 1.70000000e+02 8.69100690e-01 1.70000000e+02
1.00000000e+00 8.95043790e-01]
[5.04454970e-03 1.70000000e+02 9.08646226e-01 1.70000000e+02
1.00000000e+00 8.94312859e-01]
[7.21622109e-02 1.70000000e+02 8.15687418e-01 1.70000000e+02
1.00000000e+00 8.92976820e-01]
[4.61598933e-02 1.70000000e+02 8.43871117e-01 1.70000000e+02
1.00000000e+00 8.85869563e-01]
[8.15064907e-02 1.70000000e+02 5.09306908e-01 1.70000000e+02
1.00000000e+00 8.84744465e-01]
[3.13311994e-01 1.70000000e+02 8.06920052e-01 1.70000000e+02
1.00000000e+00 8.83312881e-01]
[5.77806830e-02 1.70000000e+02 6.74421549e-01 1.70000000e+02
1.00000000e+00 8.83202076e-01]
[7.35052526e-02 1.70000000e+02 7.99772024e-01 1.70000000e+02
1.00000000e+00 8.82957578e-01]
[9.44075286e-02 1.70000000e+02 6.36978388e-01 1.70000000e+02
1.00000000e+00 8.79572034e-01]
[1.04708567e-01 1.70000000e+02 5.06934226e-01 1.70000000e+02
1.00000000e+00 8.78465176e-01]
[3.09628636e-01 1.70000000e+02 7.21402168e-01 1.70000000e+02
1.00000000e+00 8.71768117e-01]
[4.17731255e-01 1.70000000e+02 7.74871826e-01 1.70000000e+02
1.00000000e+00 8.71591568e-01]
[1.02968812e-01 1.70000000e+02 8.31885815e-01 1.70000000e+02
1.00000000e+00 8.70697320e-01]
[3.26686859e-01 1.70000000e+02 9.03913379e-01 1.70000000e+02
1.00000000e+00 8.65643442e-01]
[2.11390853e-01 1.70000000e+02 7.61673510e-01 1.70000000e+02
1.00000000e+00 8.44844162e-01]
[6.40451163e-02 1.70000000e+02 4.58926082e-01 1.70000000e+02
1.00000000e+00 8.42892468e-01]
[8.91592205e-02 1.70000000e+02 6.34816885e-01 1.70000000e+02
1.00000000e+00 8.42469156e-01]
[8.19413960e-02 1.70000000e+02 8.82329106e-01 1.70000000e+02
1.00000000e+00 8.39045584e-01]
[1.20993361e-01 1.70000000e+02 4.17110443e-01 1.70000000e+02
1.00000000e+00 8.37420464e-01]
[7.96370357e-02 1.70000000e+02 3.95659626e-01 1.70000000e+02
1.00000000e+00 8.33381474e-01]
[5.12502432e-01 1.70000000e+02 8.35814953e-01 1.70000000e+02
1.00000000e+00 8.32887769e-01]
[4.17931348e-01 1.70000000e+02 9.30028081e-01 1.70000000e+02
1.00000000e+00 8.31515372e-01]
[1.30817384e-01 1.70000000e+02 5.15310645e-01 1.70000000e+02
1.00000000e+00 8.27732623e-01]
[2.68170387e-01 1.70000000e+02 7.14416146e-01 1.70000000e+02
1.00000000e+00 8.26003551e-01]
[5.26177883e-02 1.70000000e+02 4.45707768e-01 1.70000000e+02
1.00000000e+00 8.24885964e-01]
[2.30032802e-01 1.70000000e+02 8.43989849e-01 1.70000000e+02
1.00000000e+00 8.22971344e-01]
[2.30556309e-01 1.70000000e+02 8.52416933e-01 1.70000000e+02
1.00000000e+00 8.17419052e-01]
[3.39487523e-01 1.70000000e+02 9.32341218e-01 1.70000000e+02
1.00000000e+00 8.15800905e-01]
[8.40027183e-02 1.70000000e+02 4.93051171e-01 1.70000000e+02
1.00000000e+00 8.13550532e-01]
[3.68743122e-01 1.70000000e+02 7.81924725e-01 1.70000000e+02
1.00000000e+00 8.12513292e-01]
[3.48516613e-01 1.70000000e+02 9.05591726e-01 1.70000000e+02
1.00000000e+00 8.08090568e-01]
[1.08621031e-01 1.70000000e+02 5.89192033e-01 1.70000000e+02
1.00000000e+00 8.03775489e-01]
[1.04770765e-01 1.70000000e+02 4.27024066e-01 1.70000000e+02
1.00000000e+00 8.00240040e-01]
[3.22164953e-01 1.70000000e+02 7.51090050e-01 1.70000000e+02
1.00000000e+00 7.98988163e-01]
[9.63819027e-03 1.70000000e+02 8.10012162e-01 1.70000000e+02
1.00000000e+00 7.96526074e-01]
[6.94726408e-02 1.70000000e+02 6.51571989e-01 1.70000000e+02
1.00000000e+00 7.83412933e-01]
[1.59009308e-01 1.70000000e+02 4.58558708e-01 1.70000000e+02
1.00000000e+00 7.72169471e-01]
[9.69279855e-02 1.70000000e+02 4.14826453e-01 1.70000000e+02
1.00000000e+00 7.36888528e-01]
[9.41606164e-02 1.70000000e+02 5.42486250e-01 1.70000000e+02
1.00000000e+00 7.35704958e-01]
[1.19749263e-01 1.70000000e+02 4.83192563e-01 1.70000000e+02
1.00000000e+00 7.33437300e-01]
[1.34951457e-01 1.70000000e+02 5.17964125e-01 1.70000000e+02
1.00000000e+00 7.32905149e-01]
[0.00000000e+00 1.70000000e+02 9.54772532e-01 1.70000000e+02
1.00000000e+00 7.25317955e-01]
[2.65806049e-01 1.70000000e+02 9.20489073e-01 1.70000000e+02
1.00000000e+00 7.22334146e-01]
[1.56688362e-01 1.70000000e+02 6.60517335e-01 1.70000000e+02
1.00000000e+00 7.08233118e-01]]]

预测多目标？

你好，我想问一下这个模型可以检测出同一张图片里的多个目标吗，因为mask-RCNN理论上是可以测出多目标的