[WIP] Continue work on Integration with DeepLabCut 3.0 #127

maximpavliv · 2025-06-16T10:05:34Z

This PR continues the work started in #121, which was left incomplete.

Specifically:

It merges the main branch, thus solving the current conflict in [WIP] Integration with DeepLabCut 3.0 - PyTorch Engine #121 with main
It merges the 3 benchmarking modules currently present in [WIP] Integration with DeepLabCut 3.0 - PyTorch Engine #121 (dlclive/benchmark.py, dlclive/benchmark_tf.py, dlclive/benchmark_pytorch.py) into a single dlclive/benchmark.py module, that works with models of both backends (TF and Torch).

Remaining work:

get_system_info() in dlclive/benchmark.py uses torch to get GPU data. We want an implementation that uses the library corresponding to the model in use (tf or torch).
Add pytorch models to dlclive/check_install/check_install.py , benchmarking/run_dlclive_benchmark.py, and tests/test_benchmark_script.py - once pytorch models have been selected and uploaded online!

Then, #121 will be ready for merge into main

- changed benchmarking link; needs testing

* Update .gitignore * CI/CD update python version * CI/CD update actions versions * CI/CD update trigger events * CI/CD update MacOS version * dlclibrary set version to >=0.0.6 * Poetry lock * Poetry lock * Pyproject.toml update tensorflow installation * Poetry lock * Install specific tensorflow-io-gcs-filesystem for windows * Poetry lock * Update deprecated section name * Poetry lock * CI/CD test on python 3.11, 3.12, 3.13 as well * Update testing.yml - rename to main - add tables installation --------- Co-authored-by: Mackenzie Mathis <mathis@rowland.harvard.edu>

* Update README.md fix test instructions * Update check_install.py * Update check_install.py * Update check_install.py

* Update pyproject.toml - adding support for python 3.11, 3.12, 3.13 * Update testing.yml * Update testing.yml * Update pyproject.toml * Update testing.yml * Update pyproject.toml * Update poetry.lock * Update poetry.lock * Update pyproject.toml * Update pyproject.toml * Update pyproject.toml * Update pyproject.toml * Update pyproject.toml * Update pyproject.toml * Update pyproject.toml * Update pyproject.toml * Update pyproject.toml * Update testing.yml * Update pyproject.toml * Update pyproject.toml * Update pyproject.toml * Update testing.yml * Update pyproject.toml * Update pyproject.toml * Update poetry.lock * Update testing.yml * Update testing.yml * Update testing.yml - remove poetry (for now) * Update testing.yml * Update testing.yml * Update testing.yml * Update testing.yml * Update testing.yml * updated poetry.lock * updated * Update testing.yml * Update pyproject.toml * Update pyproject.toml * Update pyproject.toml * Update pyproject.toml * Update testing.yml testing 3.11 * Update testing.yml * Update pyproject.toml * Update pyproject.toml * Update pyproject.toml * Update testing.yml * updated test for win32 * Update testing.yml * Update testing.yml * Update testing.yml * Update testing.yml

…ence (#138)

Copilot

Pull request overview

This PR continues the integration of DeepLabCut 3.0 with PyTorch support, merging the main branch and consolidating three separate benchmarking modules into a single unified module that works with both TensorFlow and PyTorch backends.

Changes:

Merged benchmark_tf.py and benchmark_pytorch.py into a single benchmark.py module supporting both engines
Added Engine enum class to detect and classify model types (TensorFlow vs PyTorch)
Updated factory pattern and check_install scripts to use the new Engine detection
Fixed PyTorch runner bug where empty detection batches caused issues

Reviewed changes

Copilot reviewed 15 out of 16 changed files in this pull request and generated 27 comments.

Show a summary per file

File	Description
dlclive/engine.py	New Engine enum for TensorFlow/PyTorch model type detection
dlclive/benchmark.py	Unified benchmarking module supporting both TF and PyTorch
dlclive/benchmark_tf.py	Removed - merged into benchmark.py
dlclive/benchmark_pytorch.py	Removed - merged into benchmark.py
dlclive/factory.py	Updated to use Engine enum for model type detection
dlclive/pose_estimation_pytorch/runner.py	Fixed empty detection batch handling
dlclive/check_install/check_install.py	Updated to support both engine types
dlclive/init.py	Added exports for benchmark functions
tests/test_benchmark_script.py	New functional test for benchmarking
pytest.ini	Added pytest configuration with functional marker
pyproject.toml	Updated dependencies, moved opencv-python-headless, added pytest
benchmarking/run_dlclive_benchmark.py	Updated to use Engine detection
README.md	Updated test instructions
.gitignore	Added __MACOSX exclusion
.github/workflows/testing.yml	Updated CI to support conda and both engines
.github/workflows/python-package.yml	Updated branch references from master to main

Comments suppressed due to low confidence (2)

dlclive/benchmark.py:868

Keyword argument 'precision' is not a supported parameter name of function benchmark_videos.
Keyword argument 'save_dir' is not a supported parameter name of function benchmark_videos.
Keyword argument 'device' is not a supported parameter name of function benchmark_videos.
Keyword argument 'draw_keypoint_names' is not a supported parameter name of function benchmark_videos.
Keyword argument 'get_sys_info' is not a supported parameter name of function benchmark_videos.

    benchmark_videos(
        video_path=args.video_path,
        model_path=args.model_path,
        model_type=args.model_type,
        device=args.device,
        precision=args.precision,
        display=args.display,
        pcutoff=args.pcutoff,
        display_radius=args.display_radius,
        resize=tuple(args.resize) if args.resize else None,
        cropping=args.cropping,
        dynamic=tuple(args.dynamic),
        save_poses=args.save_poses,
        save_dir=args.save_dir,
        draw_keypoint_names=args.draw_keypoint_names,
        cmap=args.cmap,
        get_sys_info=args.get_sys_info,
        save_video=args.save_video,
    )

dlclive/check_install/check_install.py:13

This import of module urllib.request is redundant, as it was previously imported on line 9.

import urllib.request

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-01-13T12:02:50Z

pyproject.toml

 pandas = ">=1.0.1,!=1.5.0"
 tables = "^3.8"
-opencv-python-headless = "^4.5"
+pytest = "^8.0"


Adding pytest as a regular dependency (not a dev dependency) is unusual. Test frameworks are typically dev dependencies. Consider moving pytest to [tool.poetry.group.dev.dependencies] instead.

Copilot · 2026-01-13T12:02:51Z

tests/test_benchmark_script.py

+    dog_models = glob.glob(str(datafolder / "dog" / "*[!avi]"))
+    dog_video = glob.glob(str(datafolder / "dog" / "*.avi"))[0]
+    mouse_models = glob.glob(str(datafolder / "mouse_lick" / "*[!avi]"))


The glob pattern "*[!avi]" doesn't correctly exclude .avi files. This pattern matches any path that doesn't end with 'i' (just the last character). To properly exclude .avi files, use a more explicit approach like filtering the results or using a different pattern like "*/[!.]*.pb" for TensorFlow models or "*/*.pt" for PyTorch models.

Suggested change

dog_models = glob.glob(str(datafolder / "dog" / "*[!avi]"))

dog_video = glob.glob(str(datafolder / "dog" / "*.avi"))[0]

mouse_models = glob.glob(str(datafolder / "mouse_lick" / "*[!avi]"))

dog_models = [p for p in glob.glob(str(datafolder / "dog" / "*")) if not p.lower().endswith(".avi")]

dog_video = glob.glob(str(datafolder / "dog" / "*.avi"))[0]

mouse_models = [p for p in glob.glob(str(datafolder / "mouse_lick" / "*")) if not p.lower().endswith(".avi")]

Copilot · 2026-01-13T12:02:51Z

tests/test_benchmark_script.py

+        result = benchmark_videos(
+            model_path=model_path,
+            model_type="base" if Engine.from_model_path(model_path) == Engine.TENSORFLOW else "pytorch",
+            video_path=dog_video,
+            output=str(out_dir),
+            n_frames=n_frames,
+            pixels=pixels
+        )
+        print("Dog model result:", result)


The benchmark_videos() function doesn't return a value according to its implementation (it only saves results to disk when output is provided), but the test stores and prints the return value. This could lead to confusion. Either remove the assignment and print statements, or check the actual output files instead.

Copilot · 2026-01-13T12:02:51Z

dlclive/check_install/check_install.py


-    # make temporary directory in $HOME
-    # TODO: why create this temp directory in $HOME?
+    # make temporary directory in $current


The comment says "make temporary directory in $current" but the code actually creates it in the home directory (Path().home() / "dlc-live-tmp"). Update the comment to accurately reflect where the directory is created, or change the code to match the comment.

Suggested change

# make temporary directory in $current

# make temporary directory in the user's home directory

Copilot · 2026-01-13T12:02:52Z

dlclive/benchmark.py

+    video_path = video_path if type(video_path) is list else [video_path]
+
+    # fix resize
+    if pixels:
+        pixels = pixels if type(pixels) is list else [pixels]
+        resize = [None for p in pixels]
+    elif resize:
+        resize = resize if type(resize) is list else [resize]


Using type(pixels) is list and type(resize) is list are not recommended Python practices. Use isinstance() instead for type checking.

Suggested change

video_path = video_path if type(video_path) is list else [video_path]

# fix resize

if pixels:

pixels = pixels if type(pixels) is list else [pixels]

resize = [None for p in pixels]

elif resize:

resize = resize if type(resize) is list else [resize]

video_path = video_path if isinstance(video_path, list) else [video_path]

# fix resize

if pixels:

pixels = pixels if isinstance(pixels, list) else [pixels]

resize = [None for p in pixels]

elif resize:

resize = resize if isinstance(resize, list) else [resize]

Copilot · 2026-01-13T12:02:58Z

tests/test_benchmark_script.py

@@ -0,0 +1,48 @@
+import os
+import glob
+import pathlib


Import of 'pathlib' is not used.

Suggested change

import pathlib

Copilot · 2026-01-13T12:02:58Z

dlclive/benchmark.py

+    """
+    Downloads and extracts DeepLabCut-Live benchmarking data (videos & DLC models).
+    """
+    import os


This import of module os is redundant, as it was previously imported on line 757.

Suggested change

import os

Copilot · 2026-01-13T12:02:59Z

dlclive/benchmark.py

+    except:
+        fourcc = ""
+    try:
+        fps = round(cap.get(cv2.CAP_PROP_FPS))
+    except:
+        fps = None
+    try:
+        pix_fmt = decode_fourcc(cap.get(cv2.CAP_PROP_CODEC_PIXEL_FORMAT))
+    except:
+        pix_fmt = ""
+    try:
+        frame_count = round(cap.get(cv2.CAP_PROP_FRAME_COUNT))
+    except:
+        frame_count = None
+    try:
+        orig_im_size = (
+            round(cap.get(cv2.CAP_PROP_FRAME_WIDTH)),
+            round(cap.get(cv2.CAP_PROP_FRAME_HEIGHT)),
+        )
+    except:


Except block directly handles BaseException.

Suggested change

except:

fourcc = ""

try:

fps = round(cap.get(cv2.CAP_PROP_FPS))

except:

fps = None

try:

pix_fmt = decode_fourcc(cap.get(cv2.CAP_PROP_CODEC_PIXEL_FORMAT))

except:

pix_fmt = ""

try:

frame_count = round(cap.get(cv2.CAP_PROP_FRAME_COUNT))

except:

frame_count = None

try:

orig_im_size = (

round(cap.get(cv2.CAP_PROP_FRAME_WIDTH)),

round(cap.get(cv2.CAP_PROP_FRAME_HEIGHT)),

)

except:

except Exception:

fourcc = ""

try:

fps = round(cap.get(cv2.CAP_PROP_FPS))

except Exception:

fps = None

try:

pix_fmt = decode_fourcc(cap.get(cv2.CAP_PROP_CODEC_PIXEL_FORMAT))

except Exception:

pix_fmt = ""

try:

frame_count = round(cap.get(cv2.CAP_PROP_FRAME_COUNT))

except Exception:

frame_count = None

try:

orig_im_size = (

round(cap.get(cv2.CAP_PROP_FRAME_WIDTH)),

round(cap.get(cv2.CAP_PROP_FRAME_HEIGHT)),

)

except Exception:

Copilot · 2026-01-13T12:02:59Z

dlclive/benchmark.py

+    except:
+        fourcc = ""
+    try:
+        fps = round(cap.get(cv2.CAP_PROP_FPS))
+    except:
+        fps = None
+    try:
+        pix_fmt = decode_fourcc(cap.get(cv2.CAP_PROP_CODEC_PIXEL_FORMAT))
+    except:
+        pix_fmt = ""
+    try:
+        frame_count = round(cap.get(cv2.CAP_PROP_FRAME_COUNT))
+    except:
+        frame_count = None
+    try:
+        orig_im_size = (
+            round(cap.get(cv2.CAP_PROP_FRAME_WIDTH)),
+            round(cap.get(cv2.CAP_PROP_FRAME_HEIGHT)),
+        )
+    except:


Except block directly handles BaseException.

Suggested change

except:

fourcc = ""

try:

fps = round(cap.get(cv2.CAP_PROP_FPS))

except:

fps = None

try:

pix_fmt = decode_fourcc(cap.get(cv2.CAP_PROP_CODEC_PIXEL_FORMAT))

except:

pix_fmt = ""

try:

frame_count = round(cap.get(cv2.CAP_PROP_FRAME_COUNT))

except:

frame_count = None

try:

orig_im_size = (

round(cap.get(cv2.CAP_PROP_FRAME_WIDTH)),

round(cap.get(cv2.CAP_PROP_FRAME_HEIGHT)),

)

except:

except Exception:

fourcc = ""

try:

fps = round(cap.get(cv2.CAP_PROP_FPS))

except Exception:

fps = None

try:

pix_fmt = decode_fourcc(cap.get(cv2.CAP_PROP_CODEC_PIXEL_FORMAT))

except Exception:

pix_fmt = ""

try:

frame_count = round(cap.get(cv2.CAP_PROP_FRAME_COUNT))

except Exception:

frame_count = None

try:

orig_im_size = (

round(cap.get(cv2.CAP_PROP_FRAME_WIDTH)),

round(cap.get(cv2.CAP_PROP_FRAME_HEIGHT)),

)

except Exception:

Copilot · 2026-01-13T12:02:59Z

dlclive/benchmark.py


-    return inf_times, im_size, meta
+    os.makedirs(os.path.normpath(output), exist_ok=True)
+    pickle.dump(data, open(out_file, "wb"))


File is opened but is not closed.

Suggested change

pickle.dump(data, open(out_file, "wb"))

with open(out_file, "wb") as f:

pickle.dump(data, f)

MMathisLab and others added 16 commits June 13, 2025 16:15

Update benchmark.py

143f3d3

- changed benchmarking link; needs testing

fix #112 - don't subset cpuinfo (#114)

41ee58f

Merge branch 'main' into MMathisLab-data

ea8ce75

master --> main

36557a8

Merge branch 'main' into MMathisLab-data

6cae578

Update testing.yml

4004ef9

Update benchmark.py

f8fc8ba

Merge branch 'main' into MMathisLab-data

ae95d0b

Update README & benchmarking script (#126)

88038ac

* Update README.md fix test instructions * Update check_install.py * Update check_install.py * Update check_install.py

Merge branch 'main' into MMathisLab-data

3b88f25

Merge branch 'main' into MMathisLab-data

0883c17

Fix download_benchmarking_data()

6c61b29

Ignore __MACOSX directories

ef37746

Merge branch 'main' into maxim/dlclive3

2a4f4af

maximpavliv changed the title ~~[WIP] Continue work on _Integration with DeepLabCut 3.0_~~ [WIP] Continue work on Integration with DeepLabCut 3.0 Jun 16, 2025

maximpavliv added 3 commits June 16, 2025 13:12

Run tests on all PRs

ad14faa

tensorflow-io-gcs-filesystem - python=3.9

d882b6b

Pin python to 3.10 version

7599121

maximpavliv force-pushed the maxim/dlclive3 branch from 0cc7137 to 7599121 Compare June 16, 2025 13:08

maximpavliv and others added 9 commits June 16, 2025 15:47

Merge branch 'MMathisLab-data' into maxim/dlclive3

7593c87

WIP refactor benchmarking: __init__.py

199adc0

WIP refactor benchmarking: remove dupplicates

e86db2d

WIP refactor benchmarking: warning

0d6b2d5

WIP refactor benchmarking: close live

6402380

WIP refactor benchmarking: resize and pixels

f013557

WIP refactor benchmarking: get_sys_info

39c27f6

Update benchmark.py (#123)

44c960f

WIP refactor benchmarking: Delete dupplicate scripts

f2c6b11

maximpavliv added 4 commits June 17, 2025 09:03

WIP refactor benchmarking: Introduce Engine

bc036a7

WIP refactor benchmarking: benchmark() method

4bbcf7c

WIP refactor benchmarking: benchmark_videos()

8366187

CI/CD: add torch installation

db6a61d

maximpavliv force-pushed the maxim/dlclive3 branch from b3a7ae8 to db6a61d Compare June 17, 2025 08:05

maximpavliv added 2 commits June 17, 2025 10:21

WIP refactor benchmarking: extract Engine

664835c

Refactor benchmarking: model_type in benchmark_videos()

4475ea2

maximpavliv force-pushed the maxim/dlclive3 branch from e9177a9 to 09116e3 Compare June 18, 2025 07:16

Merge branch 'main' into maxim/dlclive3

e994808

maximpavliv force-pushed the maxim/dlclive3 branch from 09116e3 to e994808 Compare June 18, 2025 07:33

maximpavliv and others added 2 commits June 18, 2025 15:25

Delete poetry lock

5731ebe

Fix frame_batch initialization in runner.py when low detection confid…

ce65a5a

…ence (#138)

MMathisLab requested a review from Copilot January 13, 2026 11:53

Copilot started reviewing on behalf of MMathisLab January 13, 2026 11:54 View session

MMathisLab merged commit ac75b2a into dlclive3 Jan 13, 2026
6 of 7 checks passed

MMathisLab deleted the maxim/dlclive3 branch January 13, 2026 11:54

Copilot AI reviewed Jan 13, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[WIP] Continue work on Integration with DeepLabCut 3.0 #127

[WIP] Continue work on Integration with DeepLabCut 3.0 #127

Uh oh!

maximpavliv commented Jun 16, 2025 •

edited

Loading

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Jan 13, 2026

Uh oh!

Copilot AI Jan 13, 2026

Uh oh!

Copilot AI Jan 13, 2026

Uh oh!

Copilot AI Jan 13, 2026

Uh oh!

Copilot AI Jan 13, 2026

Uh oh!

Copilot AI Jan 13, 2026

Uh oh!

Copilot AI Jan 13, 2026

Uh oh!

Copilot AI Jan 13, 2026

Uh oh!

Copilot AI Jan 13, 2026

Uh oh!

Copilot AI Jan 13, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

	# make temporary directory in $current
	# make temporary directory in the user's home directory

	pickle.dump(data, open(out_file, "wb"))
	with open(out_file, "wb") as f:
	pickle.dump(data, f)

[WIP] Continue work on Integration with DeepLabCut 3.0 #127

[WIP] Continue work on Integration with DeepLabCut 3.0 #127

Uh oh!

Conversation

maximpavliv commented Jun 16, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

maximpavliv commented Jun 16, 2025 •

edited

Loading