torch / demos

Demos and tutorials around Torch7.
355 stars 301 forks source link

person-detector camera error #66

Open kaisark opened 6 years ago

kaisark commented 6 years ago

I was able to run qlua run.lua --visualize successfully, but the rundemo.lua is failing. Initially, I ran into a deprecated issue and commented out the require 'torchffi' line. I'm not sure if I am running into what might be a camera (Sony Playstation Eye) related issue running the demo (see below).

nvidia@tegra-ubuntu:~/demos/person-detector$ qlua rundemo.lua -c 0 ==> processing options Neural Network used: nn.Sequential { [input -> (1) -> (2) -> (3) -> (4) -> (5) -> (6) -> (7) -> (8) -> output] (1): nn.SpatialConvolutionMM(3 -> 32, 7x7, 1,1) (2): nn.Threshold (3): nn.SpatialMaxPooling(2x2, 2,2) (4): nn.SpatialConvolutionMM(32 -> 64, 7x7, 1,1) (5): nn.Threshold (6): nn.SpatialMaxPooling(2x2, 2,2) (7): nn.SpatialConvolutionMM(64 -> 128, 7x7, 1,1) (8): nn.SpatialClassifier } Initializing device: /dev/video0 FPS wanted 1 Using 1 buffers Warning: camera resolution changed to 240x320 camera[0] started : 1 ./PyramidUnPacker.lua:97: bad argument #2 to 'narrow' (out of range at /home/nvidia/torch/pkg/torch/lib/TH/generic/THTensor.c:438) stack traceback: [C]: in function 'narrow' ./PyramidUnPacker.lua:97: in function 'forward' rundemo.lua:129: in function 'process' rundemo.lua:183: in function