SAM and GroundingDINO failures: separate code, checkpoints and extensions

Identify the implementation first

The original Segment Anything project documents code installation separately from checkpoints. GroundingDINO has its own installation, configuration and weight requirements.12 Similar names do not establish interchangeability between SAM, SAM2 or third-party wrappers.

Our suggested checklist records the calling node revision, exact import, configuration and checkpoint origin. A missing Python module is an installation-path question. An _C or other native-extension failure needs its compilation or loading error preserved. Changing a model path is not a substitute for that investigation.

python -m pip show torch torchvision
python -c "import sys, platform; print(sys.version); print(platform.machine())"

Some node projects vendor code, so the absence of a distribution record alone is inconclusive. Read the node's own integration instructions.

Test the failing segmentation or detection stage on a small non-private image. Check the selected configuration and checkpoint, then the mask dimensions and downstream connections. Change one component at a time. Where platform support is unconfirmed, record the uncertainty instead of using an unidentified wheel or presenting somebody else's screenshot as your test.

Related checks

• Fix cv2 imports without stacking OpenCV distributions
• MMCV installation and missing-operator errors: check the build combination

Research draft. No local/GPU reproduction or independent editorial sign-off. Upstream statements and our recommendations are distinguished.