search

mzolfaghari / ECO-efficient-video-understanding

Code and models of paper " ECO: Efficient Convolutional Network for Online Video Understanding", ECCV 2018
MIT License
437 stars 96 forks source link

About scripts for online recognition and video captioning #24

Closed RaRaRan closed 5 years ago

RaRaRan commented 5 years ago

Hi ,I am very interested in your outstanding work. But I have a problem when i run online recognition.py I set the model paths in online_recognition.py as :

rgb_mean.mat: /home/ra/ECO-efficient-video-understanding-master/caffe_3d/action_matlab deploy.prototxt: /home/ra/ECO-efficient-video-understanding-master/models_ECO_Lite/kinetics ECO_Lite_kinetics.caffemodel: download from Google Drive

And i got error: python online_recognition.py Setting device 0 WARNING: Logging before InitGoogleLogging() is written to STDERR I1105 17:18:41.087724 7888 net.cpp:46] Initializing net from parameters: name: "ECOLite" input: "data" input_dim: 80 input_dim: 3 input_dim: 224 input_dim: 224 . . . I1105 17:19:27.531430 7888 net.cpp:551] Collecting Learning Rate and Weight Decay. I1105 17:19:27.531447 7888 net.cpp:300] Network initialization done. I1105 17:19:27.531451 7888 net.cpp:301] Memory required for data: 3935844160 GLib-GIO-Message: 17:19:43.243: Using the 'memory' GSettings backend. Your settings will not be saved or shared with other applications. [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15] Traceback (most recent call last): File "online_recognition.py", line 122, in online_predict(mean_file,model_def_file,model_file,classes_file,num_categories) File "online_recognition.py", line 92, in online_predict net.blobs['data'].data[...] = np.transpose(rgb[:,:,:,:], (3,2,1,0)) ValueError: could not broadcast input array from shape (16,3,224,224) into shape (80,3,224,224)

Did I set it wrong? Please help me,thanks!

RaRaRan commented 5 years ago

solved! i set input_dim:80 to 16. done!