diff --git a/deps/build.jl b/deps/build.jl
index 8b4c254cb..c0042f4fc 100644
--- a/deps/build.jl
+++ b/deps/build.jl
@@ -68,6 +68,14 @@ else
   info("Did not find a CUDA installation, using CPU-only version of MXNet.")
 end
 
+# try to find opencv
+HAS_OPENCV = false
+let libopencv_core = Libdl.find_library("libopencv_core")
+  HAS_OPENCV = !isempty(libopencv_core) && Libdl.dlopen_e(libopencv_core) != C_NULL
+end
+
+info("Build with OpenCV -> $(HAS_OPENCV)")
+
 function get_cpucore()
     if haskey(ENV, "TRAVIS")  # on travis-ci
         2
@@ -139,7 +147,7 @@ if !libmxnet_detected
     USE_JULIA_BLAS = true
     FORCE_LAPACK = true
   end
-  info("USE_JULIA_BLAS -> $USE_JULIA_BLAS")
+  info("Build with Julia's BLAS lib -> $USE_JULIA_BLAS")
 
   blas_name = blas_vendor == :openblas64 ? "openblas" : string(blas_vendor)
   MSHADOW_LDFLAGS = "MSHADOW_LDFLAGS=-lm $blas_path"
@@ -188,7 +196,9 @@ if !libmxnet_detected
           end
 
           # Configure OpenCV
-          `sed -i -s 's/USE_OPENCV = 1/USE_OPENCV = 0/' config.mk`
+          if !HAS_OPENCV
+            `sed -i -s 's/USE_OPENCV = 1/USE_OPENCV = 0/' config.mk`
+          end
 
           # Configure CUDA
           if HAS_CUDA
diff --git a/docs/src/user-guide/install.md b/docs/src/user-guide/install.md
index 30ed65d48..7160d387a 100644
--- a/docs/src/user-guide/install.md
+++ b/docs/src/user-guide/install.md
@@ -21,19 +21,42 @@ MXNet.jl is built on top of [libmxnet](https://github.com/dmlc/mxnet).
 Upon installation, Julia will try to automatically download and build
 libmxnet.
 
-There are three environment variables that change this behaviour. If you
-already have a pre-installed version of mxnet you can use `MXNET_HOME`
-to point the build-process in the right direction. If the automatic
-cuda detection fails you can also set `CUDA_HOME` to override the process.
-To control which version of libmxnet will be compiled, you can use the
-`MXNET_COMMIT` variable to point to either a version tag (e.g. `v0.10.0`), a
-branch name (e.g. `master`) or a specific commit hash (e.g. `a0b1c2d3`).
+There are three environment variables that change this behaviour.
+
+- `MXNET_HOME`: If you already have a pre-installed version of mxnet,
+  you can use `MXNET_HOME` to point the build-process in the right direction.
+
+- `CUDA_HOME`: If the automatic cuda detection fails,
+  you can also set `CUDA_HOME` to override the process.
+
+  ```julia
+  ENV["CUDA_HOME"] = "/usr/mycuda"
+  Pkg.build("MXNet")
+  ```
+
+- `MXNET_COMMIT`: To control which version of libmxnet will be compiled,
+  you can use the `MXNET_COMMIT` variable to point to either a version tag
+  (e.g. `v0.10.0`), a branch name (e.g. `master`) or a specific commit hash
+  (e.g. `a0b1c2d3`).
+
+  ```julia
+  ENV["MXNET_COMMIT"] = "master"
+  Pkg.build("MXNet")
+  ```
 
 The libmxnet source is downloaded to `Pkg.dir("MXNet", "deps", "src", "mxnet")`.
-The automatic build is using default configurations, with OpenCV disabled.
-If the compilation failed due to unresolved dependency, or if
-you want to customize the build, you can compile and
-install libmxnet manually. Please see below for more details.
+The automatic build will try to detect optional dependencies, including
+
+- [CUDA](https://developer.nvidia.com/cuda-toolkit) 8.0
+
+- [CuDNN](https://developer.nvidia.com/cudnn)
+
+- [OpenCV](https://opencv.org/)
+
+If the compilation failed due to unresolved dependency,
+or if you want to customize the build,
+you can compile and install libmxnet manually.
+Please see below for more details.
 
 Manual Compilation
 ------------------