Merge pull request tesseract-ocr#2461 from stweil/tensorflow

Support build with Tensorflow
danrufus · May 25, 2019 · 8de022a · 8de022a
2 parents e44c60c + 32dcfd0
commit 8de022a
Show file tree

Hide file tree

Showing 13 changed files with 1,725 additions and 12 deletions.
diff --git a/configure.ac b/configure.ac
@@ -195,6 +195,25 @@ if test "$enable_opencl" = "yes"; then
   ])
 fi
 
+# Check whether to build with support for TensorFlow.
+AC_ARG_WITH([tensorflow],
+  AS_HELP_STRING([--with-tensorflow],
+                 [support TensorFlow @<:@default=check@:>@]),
+  [], [with_tensorflow=check])
+AM_CONDITIONAL([TENSORFLOW], false)
+TENSORFLOW_LIBS=
+AS_IF([test "x$with_tensorflow" != xno],
+  [AC_CHECK_FILE([/usr/include/tensorflow/core/framework/graph.pb.h],
+    [AC_SUBST([TENSORFLOW_LIBS], ["-lprotobuf -ltensorflow_cc"])
+      AM_CONDITIONAL([TENSORFLOW], true)
+    ],
+    [if test "x$with_tensorflow" != xcheck; then
+      AC_MSG_FAILURE(
+        [--with-tensorflow was given, but test for libtensorflow-dev failed])
+     fi
+    ])
+  ])
+
 # https://lists.apple.com/archives/unix-porting/2009/Jan/msg00026.html
 m4_define([MY_CHECK_FRAMEWORK],
   [AC_CACHE_CHECK([if -framework $1 works],[my_cv_framework_$1],

diff --git a/src/api/Makefile.am b/src/api/Makefile.am
@@ -88,6 +88,7 @@ tesseract_LDFLAGS = $(OPENCL_LDFLAGS)
 
 tesseract_LDADD += $(LEPTONICA_LIBS)
 tesseract_LDADD += $(OPENMP_CXXFLAGS)
+tesseract_LDADD += $(TENSORFLOW_LIBS)
 tesseract_LDADD += $(libarchive_LIBS)
 
 if T_WIN

diff --git a/src/ccutil/unicharset.h b/src/ccutil/unicharset.h
@@ -2,7 +2,6 @@
 // File:        unicharset.h
 // Description: Unicode character/ligature set class.
 // Author:      Thomas Kielbus
-// Created:     Wed Jun 28 17:05:01 PDT 2006
 //
 // (C) Copyright 2006, Google Inc.
 // Licensed under the Apache License, Version 2.0 (the "License");
@@ -153,7 +152,7 @@ class UNICHARSET {
   // List of strings for the SpecialUnicharCodes. Keep in sync with the enum.
   static TESS_API const char* kSpecialUnicharCodes[SPECIAL_UNICHAR_CODES_COUNT];
 
-  // ICU 2.0 UCharDirection enum (from third_party/icu/include/unicode/uchar.h)
+  // ICU 2.0 UCharDirection enum (from icu/include/unicode/uchar.h)
   enum Direction {
       U_LEFT_TO_RIGHT               = 0,
       U_RIGHT_TO_LEFT               = 1,

diff --git a/src/lstm/Makefile.am b/src/lstm/Makefile.am
@@ -10,6 +10,11 @@ AM_CPPFLAGS += \
 
 AM_CXXFLAGS = $(OPENMP_CXXFLAGS)
 
+if TENSORFLOW
+AM_CPPFLAGS += -DINCLUDE_TENSORFLOW
+AM_CPPFLAGS += -I/usr/include/tensorflow
+endif
+
 if !NO_TESSDATA_PREFIX
 AM_CXXFLAGS += -DTESSDATA_PREFIX=@datadir@
 endif
@@ -37,3 +42,7 @@ libtesseract_lstm_la_SOURCES = \
         networkbuilder.cpp network.cpp networkio.cpp \
         parallel.cpp plumbing.cpp recodebeam.cpp reconfig.cpp reversed.cpp \
         series.cpp stridemap.cpp tfnetwork.cpp weightmatrix.cpp
+
+if TENSORFLOW
+libtesseract_lstm_la_SOURCES += tfnetwork.pb.cc
+endif
diff --git a/src/lstm/tfnetwork.cpp b/src/lstm/tfnetwork.cpp
@@ -3,7 +3,6 @@
 // Description: Encapsulation of an entire tensorflow graph as a
 //              Tesseract Network.
 // Author:      Ray Smith
-// Created:     Fri Feb 26 09:35:29 PST 2016
 //
 // (C) Copyright 2016, Google Inc.
 // Licensed under the Apache License, Version 2.0 (the "License");
@@ -90,14 +89,14 @@ void TFNetwork::Forward(bool debug, const NetworkIO& input,
   if (!model_proto_.image_widths().empty()) {
     TensorShape size_shape{1};
     Tensor width_tensor(tensorflow::DT_INT64, size_shape);
-    auto eigen_wtensor = width_tensor.flat<int64>();
+    auto eigen_wtensor = width_tensor.flat<tensorflow::int64>();
     *eigen_wtensor.data() = stride_map.Size(FD_WIDTH);
     tf_inputs.emplace_back(model_proto_.image_widths(), width_tensor);
   }
   if (!model_proto_.image_heights().empty()) {
     TensorShape size_shape{1};
     Tensor height_tensor(tensorflow::DT_INT64, size_shape);
-    auto eigen_htensor = height_tensor.flat<int64>();
+    auto eigen_htensor = height_tensor.flat<tensorflow::int64>();
     *eigen_htensor.data() = stride_map.Size(FD_HEIGHT);
     tf_inputs.emplace_back(model_proto_.image_heights(), height_tensor);
   }

diff --git a/src/lstm/tfnetwork.h b/src/lstm/tfnetwork.h
@@ -27,9 +27,9 @@
 
 #include "network.h"
 #include "static_shape.h"
-#include "tfnetwork.proto.h"
-#include "third_party/tensorflow/core/framework/graph.pb.h"
-#include "third_party/tensorflow/core/public/session.h"
+#include "tfnetwork.pb.h"
+#include "tensorflow/core/framework/graph.pb.h"
+#include "tensorflow/core/public/session.h"
 
 namespace tesseract {
 
@@ -69,6 +69,18 @@ class TFNetwork : public Network {
                NetworkScratch* scratch, NetworkIO* output) override;
 
  private:
+  // Runs backward propagation of errors on the deltas line.
+  // See Network for a detailed discussion of the arguments.
+  bool Backward(bool debug, const NetworkIO& fwd_deltas,
+                NetworkScratch* scratch,
+                NetworkIO* back_deltas) override {
+    tprintf("Must override Network::DebugWeights for type %d\n", type_);
+  }
+
+  void DebugWeights() override {
+    tprintf("Must override Network::DebugWeights for type %d\n", type_);
+  }
+
   int InitFromProto();
 
   // The original network definition for reference.