Age | Commit message (Collapse) | Author | Files | Lines | |
---|---|---|---|---|---|
2014-05-26 | caffe.Net preprocessing members belong to object, not class | Evan Shelhamer | 1 | -6/+7 | |
2014-05-23 | follow-up on #443 to invert k channels (instead of 3) | Evan Shelhamer | 1 | -1/+2 | |
2014-05-23 | Correctly invert the swapping of colour channels | James Thewlis | 1 | -1/+2 | |
In the 'deprocess' method, get back the image with the original channel order by inverting the original transform, rather than reversing the tuple which is incorrect. | |||||
2014-05-22 | fix draw_net python script | Evan Shelhamer | 1 | -2/+2 | |
include caffe.draw for drawing functions. | |||||
2014-05-20 | preprocess single inputs instead of lists | Evan Shelhamer | 3 | -42/+38 | |
For compositionality and expectations. | |||||
2014-05-20 | windowed detection in python | Evan Shelhamer | 5 | -462/+262 | |
2014-05-19 | squash infuriating loop assignment bug in batching | Evan Shelhamer | 1 | -1/+1 | |
2014-05-19 | image classification in python | Evan Shelhamer | 5 | -129/+206 | |
2014-05-19 | fix padding for the last batch | Evan Shelhamer | 1 | -3/+4 | |
2014-05-19 | split drawnet into module code and script | Evan Shelhamer | 2 | -15/+25 | |
Don't run scripts in the module dir to avoid import collisions between io and caffe.io. | |||||
2014-05-19 | add caffe.io submodule for conversions, image loading and resizing | Evan Shelhamer | 3 | -12/+89 | |
2014-05-19 | fix python mean subtraction | Evan Shelhamer | 1 | -7/+9 | |
2014-05-16 | drop cute names in favor of Net.{pre,de}process() for input formatting | Evan Shelhamer | 1 | -44/+45 | |
...and refer to inputs as inputs and not images since general vectors and matrices are perfectly fine. | |||||
2014-05-16 | Net.caffeinate() and Net.decaffeinate() format/unformat lists | Evan Shelhamer | 1 | -41/+46 | |
2014-05-16 | take blob args as ndarrays and assign on the python side | Evan Shelhamer | 2 | -184/+72 | |
Take blob args and give blob returns as single ndarrays instead of lists of arrays. Assign the net blobs and diffs as needed on the python side, which reduces copies and simplifies the C++ side of the wrapper. Thanks @longjon for the suggestion. | |||||
2014-05-15 | resize to input dimensions when formatting in python | Evan Shelhamer | 1 | -0/+7 | |
2014-05-15 | replace iterator with indices for consistency | Evan Shelhamer | 1 | -6/+6 | |
2014-05-15 | python style | Evan Shelhamer | 2 | -254/+249 | |
2014-05-15 | fix accidental revert of Init() from f5c28581 | Evan Shelhamer | 1 | -2/+1 | |
2014-05-14 | batch inputs in python by forward_all() and forward_backward_all() | Evan Shelhamer | 2 | -2/+102 | |
2014-05-14 | don't squeeze blob arrays for python | Evan Shelhamer | 1 | -2/+2 | |
Preserve the non-batch dimensions of blob arrays, even for singletons. The forward() and backward() helpers take lists of ndarrays instead of a single ndarray per blob, and lists of ndarrays are likewise returned. Note that for output the blob array could actually be returned as a single ndarray instead of a list. | |||||
2014-05-14 | python forward() and backward() extract any blobs and diffs | Evan Shelhamer | 1 | -18/+32 | |
2014-05-14 | python Net.backward() helper and Net.BackwardPrefilled() | Evan Shelhamer | 2 | -2/+40 | |
2014-05-14 | bad forward/backward inputs throw exceptions instead of crashing python | Evan Shelhamer | 1 | -13/+14 | |
2014-05-14 | pycaffe Net.forward() helper | Evan Shelhamer | 1 | -0/+36 | |
Do forward pass by prefilled or packaging input + output blobs and returning a {output blob name: output list} dict. | |||||
2014-05-14 | set input preprocessing per blob in python | Evan Shelhamer | 1 | -22/+64 | |
2014-05-14 | expose input and output blob names to python as lists | Evan Shelhamer | 1 | -0/+20 | |
2014-05-13 | pycaffe comments, lint | Evan Shelhamer | 1 | -11/+11 | |
2014-05-13 | add python io getters, mean helper, and image caffeinator/decaffeinator | Evan Shelhamer | 1 | -0/+73 | |
2014-05-13 | make python wrapper mean match binaryproto dimensions | Evan Shelhamer | 1 | -0/+0 | |
ilsvrc_2012_mean.npy has dims K x H x W. Code written for the old D x D x K mean needs to be rewritten! | |||||
2014-05-13 | match existing python formatting | Evan Shelhamer | 1 | -16/+21 | |
2014-05-02 | pycaffe: allow 1d labels to be passed to set_input_arrays | Jonathan L Long | 2 | -13/+24 | |
2014-05-02 | pycaffe: add Net.set_input_arrays for input from numpy | Jonathan L Long | 1 | -1/+65 | |
This requires a net whose first layer is a MemoryDataLayer. | |||||
2014-05-02 | pycaffe: store a shared_ptr<CaffeNet> in SGDSolver | Jonathan L Long | 1 | -1/+5 | |
Doing this, rather than constructing the CaffeNet wrapper every time, will allow the wrapper to hold references that last at least as long as SGDSolver (which will be necessary to ensure that data used by MemoryDataLayer doesn't get freed). | |||||
2014-05-02 | pycaffe: let boost pass shared_ptr<CaffeNet> | Jonathan L Long | 1 | -1/+1 | |
2014-04-25 | pycaffe: add unary CaffeNet constructor for uninitialized nets | Jonathan L Long | 1 | -3/+15 | |
Note that parameters are uninitialized, not zero-filled. | |||||
2014-04-25 | note pydot dependency of model visualization | Evan Shelhamer | 1 | -1/+6 | |
2014-04-25 | Update the drawnet.py to reflect the recent revised net definition. | ZhiHeng NIU | 1 | -4/+11 | |
2014-04-11 | fix lint errors by adding 'explicit' to new single arg pycaffe | Jeff Donahue | 1 | -2/+2 | |
constructors | |||||
2014-04-04 | pycaffe: expose SGDSolver.solve | Jonathan L Long | 1 | -1/+8 | |
2014-04-04 | pycaffe: introduce CheckFile helper | Jonathan L Long | 1 | -23/+15 | |
2014-04-04 | pycaffe: expose SGDSolver.net | Jonathan L Long | 1 | -1/+7 | |
2014-04-04 | switch from inheritance to directly overriding methods for caffe.Net | Jonathan L Long | 2 | -22/+28 | |
Also, internally drop the "Caffe" prefix for classes exported by the internal _caffe module. This change means that SGDSolver will automatically return a Net with the "nice" Python interface rather than the internal one, (and avoids the extra layer of method wrapping that results from switching to composition). | |||||
2014-04-04 | pycaffe: expose SGDSolver | Jonathan L Long | 3 | -2/+23 | |
2014-04-04 | Revert "Fix segfault." (python) | Evan Shelhamer | 1 | -1/+0 | |
Revert 88507032de1eb675539d91a1d79c221f3faf5d99 from #287. The call to `Py_Initialize()` is unnecessary and masked a build configuration issue. | |||||
2014-04-04 | Merge pull request #277 from jeffdonahue/makefile-improvements | Evan Shelhamer | 1 | -0/+0 | |
Makefile improvements | |||||
2014-04-02 | Fix segfault. | Rob Hess | 1 | -0/+1 | |
2014-04-02 | cleanup python build and clean | Jeff Donahue | 1 | -0/+0 | |
2014-03-27 | Standardize copyright, add root-level CONTRIBUTORS credit | Evan Shelhamer | 1 | -1/+1 | |
2014-03-18 | drop models/ in favor of examples/ | Evan Shelhamer | 1 | -2/+2 | |
Move the Caffe reference imagenet model and script to fetch it to examples/imagenet. Caffe bundles reference models, but it makes more sense to keep them close to examples. |