diff --git a/CHANGELOG.md b/CHANGELOG.md
index 6606198..e5483d6 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -5,7 +5,10 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 
 ## [Unreleased]
 
-## Changed
+### Fixed
+- Validate image object in do_EI is a PDFStream ([#451](https://github.com/pdfminer/pdfminer.six/pull/451))
+
+### Changed
 - Hiding fallback xref by default from dumppdf.py output ([#431](https://github.com/pdfminer/pdfminer.six/pull/431))
 
 ## [20200517]
@@ -20,6 +23,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 
 ### Added
 - Allow boxes_flow LAParam to be passed as None, validate the input, and update documentation ([#395](https://github.com/pdfminer/pdfminer.six/pull/395))
+- Also accept file-like objects in high level functions `extract_text` and `extract_pages` ([#392](https://github.com/pdfminer/pdfminer.six/pull/392))
 
 ### Fixed
 - Text no longer comes in reverse order when advanced layout analysis is disabled ([#398](https://github.com/pdfminer/pdfminer.six/pull/398))
@@ -27,9 +31,6 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 - Ignore ValueError when converting font encoding differences ([#389](https://github.com/pdfminer/pdfminer.six/pull/389))
 - Grouping of text lines outside of parent container bounding box ([#386](https://github.com/pdfminer/pdfminer.six/pull/386))
 
-### Added
-- Also accept file-like objects in high level functions `extract_text` and `extract_pages` ([#392](https://github.com/pdfminer/pdfminer.six/pull/392))
-
 ### Changed
 - Group text lines if they are centered ([#382](https://github.com/pdfminer/pdfminer.six/pull/382))
 
diff --git a/pdfminer/pdfinterp.py b/pdfminer/pdfinterp.py
index fdd54a8..ef67947 100644
--- a/pdfminer/pdfinterp.py
+++ b/pdfminer/pdfinterp.py
@@ -837,7 +837,7 @@ class PDFPageInterpreter:
 
     def do_EI(self, obj):
         """End inline image object"""
-        if 'W' in obj and 'H' in obj:
+        if isinstance(obj, PDFStream) and 'W' in obj and 'H' in obj:
             iobjid = str(id(obj))
             self.device.begin_figure(iobjid, (0, 0, 1, 1), MATRIX_IDENTITY)
             self.device.render_image(iobjid, obj)