Improve thermal camera library with community fixes#2
Open
tdamsma wants to merge 34 commits into
Open
Conversation
Added all changes to main
…operly search for interpreter
error was ``` cv2.error: OpenCV(4.10.0) 👎 error: (-5:Bad argument) in function 'inRange' > Overload resolution failed: > - upperb is not a numpy array, neither a scalar > - Expected Ptr<cv::UMat> for argument 'upperb' ```
…ix ruff findings Introduces an "AppState" as an intermediate solution for global variables. This makes access to global stuff more visible, but should vanish step by step (by more explicit dependency injection and data modelling)
Make project capable of running inside a virtual environment and format sources
…metimes returns wrong values during startup
irpythermal.py changes: - calibrate_raw() now detects both cold AND hot stuck pixels using symmetric 5% threshold (previously only detected cold pixels) Source: diminDDL/IR-Py-Thermal PR stawel#4 diminDDL#4 - Add save_calibration() and load_calibration() methods to persist reference frame and dead pixel mask as .npy files across restarts Source: MCMH2000/OpenHD_HT301_Driver noise_pattern_calibration.npy approach https://github.com/MCMH2000/OpenHD_HT301_Driver - Fix resource leak in find_device(): release cameras that don't match supported resolutions instead of leaving them open - Replace bare except with except Exception
Cast x, y, tx, ty coordinates to int() before passing to cv2.line() and cv2.putText(). OpenCV requires integer coordinates and fails with float values from numpy operations. Source: uncommitted fix found on remote machine /home/daniel/git/t2s/
Replace skimage.exposure.equalize_hist + rescale_intensity with cv2.normalize + cv2.equalizeHist. Same visual result, eliminates the scikit-image dependency (and its transitive deps) making deployment on headless Linux machines simpler.
Convert build system from poetry-core to uv/PEP 621 project format. Add missing runtime dependencies (av, opencv-python, pyusb, imageio).
Documents where this fork sits in the ht301_hacklib ecosystem, lists all known forks and their contributions, supported cameras, temperature calculation details, known calibration issues, and links to external references (papers, blog posts, teardowns).
np.kron upscaling took 32ms/frame (56% of total). cv2.resize with INTER_NEAREST gives identical pixel-doubled output but is hardware accelerated. Profiled on T2S+ 256x192 with 4x upscale: - Before: 17.6 FPS (np.kron 32ms + CLAHE 20ms) - After: ~25 FPS (cv2.resize <1ms)
calibrate_raw() with dead/hot pixel detection only applies to camera_raw=True mode (T2S+ v2 where FPGA doesn't process frames). For v1 units and HT-301, the firmware handles NUC and dead pixel correction in hardware -- only the shutter command (0x8000) is needed.
Reads from the thermal camera, applies normalize + equalize + colormap + CLAHE, and pipes BGR24 frames to /dev/video10 via ffmpeg. Replaces the duplicate scripts found at: - /home/daniel/git/t2s/frame_pusher.py (uncommitted) - /home/daniel/git/pcm/pcm/apps/pcm-capture/app/t2s_virtual.py Improvements over originals: - Uses irpythermal instead of ht301_hacklib - CLI args for device path, orientation, fps - Proper error handling and cleanup - Removes unused code (idle_frame, draw_temp, coordinate rotation)
Uses fuser to detect readers on the loopback device every 2s. Camera stays open and frames are read continuously to keep USB active, but processing and ffmpeg piping only run when needed.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Incorporates improvements from the ht301_hacklib community ecosystem into the CEAD fork.
Hot pixel detection + persistent calibration (
irpythermal.py)calibrate_raw()now detects both cold AND hot stuck pixels using a symmetric 5% threshold. Previously only cold/dead pixels were detected, leaving hot/stuck pixels uncorrected.save_calibration(path)/load_calibration(path)methods to persist reference frame and dead pixel mask as.npyfiles, surviving restarts without re-running shutter calibration.noise_pattern_calibration.npyapproachfind_device()now releases cameras that don't match supported resolutionsexcept:withexcept Exception:OpenCV int cast fix (
utils.py)int()indrawTemperature()— OpenCV rejects float coordinates from numpy/home/daniel/git/t2s/)Remove skimage dependency (
opencv.py)skimage.exposure.equalize_hist+rescale_intensitywithcv2.normalize+cv2.equalizeHistBuild system (
pyproject.toml)References
Test plan
ssh thijsd@172.16.88.136)opencv.py— verify live view with temperature overlay