All authors
Claude Skills by wenmin-wu
github.com/wenmin-wu535 skills0 installs688 views
- 2d To 3d Model ConversionConverts a pretrained 2D CNN into a 3D CNN by recursively replacing Conv2d/BN2d/Pool2d layers and inflating kernel weights along the depth axis.Votes: 0GitHub stars: 61
- 3d Center Crop Zero PadSelect center slices from a 3D volume and zero-pad along depth when the scan has fewer slices than the target countVotes: 0GitHub stars: 61
- 3d Encoder 2d Decoder Segmentation3D ResNet encoder extracts volumetric features, pools depth dimension, then feeds into a 2D UNet/FPN decoder for segmentationVotes: 0GitHub stars: 61
- 3d Mixup AugmentationApplies Mixup augmentation to 3D volumetric images and their segmentation masks, interpolating both inputs and loss targets.Votes: 0GitHub stars: 61
- 3d Patch Sliding Window InferenceTiles 3D volumes into overlapping patches for inference and averages overlapping regions to produce seamless predictions.Votes: 0GitHub stars: 61
- Albumentations Detection AugmentationApplies albumentations augmentations to object detection data while preserving bbox-label correspondence via BboxParams.Votes: 0GitHub stars: 61
- Anisotropic Spacing Trilinear ResampleResample a 3D medical volume to a fixed network input shape using physical voxel spacing (dz, dy, dx), correcting the Z dimension by the dz/dy ratio so anisotropic CT scans (1mm in-plane, 5mm slice) end up anatomically isotropic before trilinear interpolationVotes: 0GitHub stars: 61
- Arcface Metric LearningArcFace angular margin loss layer for learning discriminative embeddings — used in image retrieval, product matching, and face recognitionVotes: 0GitHub stars: 61
- Attention Lstm Image CaptioningShow-Attend-Tell decoder: Bahdanau additive attention over CNN spatial features driving a gated LSTMCell for autoregressive image-to-sequence generation.Votes: 0GitHub stars: 61
- Batch All Contrastive LossAll-vs-all contrastive loss comparing every pair in a batch (N^2 pairs) with margin and compactification regularizerVotes: 0GitHub stars: 61
- Bbox Augmentation Retry LoopWrap albumentations bbox transforms in a 10-attempt retry loop that re-rolls the augmentation when all boxes get cropped out, preventing empty-target samples from poisoning the detection loss with NaN gradientsVotes: 0GitHub stars: 61
- Bbox Interpolation Temporal TrackingInterpolate missing bounding boxes across video frames using bidirectional pandas interpolation to maintain smooth tracking through occlusionsVotes: 0GitHub stars: 61
- Bce Dice Combined LossCombines BCE-with-logits and soft Dice loss with configurable weights for binary and multilabel segmentation training.Votes: 0GitHub stars: 61
- Bev Conv1 Channel ExpansionReplace a pretrained ResNet's conv1 with a wider input conv to accept stacked BEV raster channels (semantic map + agent history) while keeping downstream weightsVotes: 0GitHub stars: 61
- Bigru Slice Feature AggregatorTwo-stage CT classifier where a 2D CNN dumps per-slice features once, then a bidirectional GRU runs over the slice sequence to produce both per-slice predictions (TimeDistributed head) and an exam-level prediction (avg+max pooled head) — turns expensive 3D CNN training into cheap sequence modelingVotes: 0GitHub stars: 61
- Black Border AutocropCrops uninformative black or dark borders from images by deriving a binary mask from the grayscale channel and trimming rows/columns.Votes: 0GitHub stars: 61
- Black Slice ReplacementDetect all-black DICOM/MRI slices (mean==0) and replace them by randomly sampling a non-black slice from the same seriesVotes: 0GitHub stars: 61
- Brovey PansharpeningFuse low-resolution multispectral bands with a high-resolution panchromatic band using the Brovey transform to produce sharp multi-band imageryVotes: 0GitHub stars: 61
- Center Z Slice SelectionSelect a fixed number of Z-slices centered around the volume midpoint for memory-efficient 2.5D input from 3D CT/MRI stacksVotes: 0GitHub stars: 61
- Centroid Distance Anomaly ScoreScore each face in a video as anomalous by computing L2 distance from the embedding centroid of all faces, then convert to probability via logistic functionVotes: 0GitHub stars: 61
- Chunked Csv Image GeneratorMemory-efficient Keras generator that streams sharded CSV files in chunks, renders strokes to images on-the-fly, and yields batches for training on datasets too large for memoryVotes: 0GitHub stars: 61
- Chunked Gpu Similarity SearchCompute pairwise cosine similarity on GPU in fixed-size chunks to avoid OOM, transferring only threshold-passing results to CPUVotes: 0GitHub stars: 61
- Class Balanced Epoch UndersamplingDynamically undersample majority class each epoch with per-class keep probabilities for stochastic balanceVotes: 0GitHub stars: 61
- Class Density Threshold Patch SamplingSample training patches from large images using per-class area-fraction thresholds to ensure each patch contains meaningful object coverageVotes: 0GitHub stars: 61
- Classifier Free Guidance DiffusionManual stable diffusion inference loop with classifier-free guidance that interpolates between unconditional and conditional noise predictions for controllable image generationVotes: 0GitHub stars: 61
- Classifier Gated Detection PostprocessUses a binary classifier's probability to gate object detector outputs in three tiers: keep detections, append a no-finding box, or replace all detections.Votes: 0GitHub stars: 61
- Clip Aesthetic ScoringScores image aesthetic quality by passing L2-normalized CLIP ViT-L/14 embeddings through a trained MLP regressor head.Votes: 0GitHub stars: 61
- Clip Interrogator CaptioningGenerate descriptive text prompts from images by combining BLIP captioning with CLIP cosine similarity against curated label banks for medium, movement, and flavor attributesVotes: 0GitHub stars: 61
- Cnn Encoder Spatial Feature MapStrips the global pool and FC head from a pretrained CNN to expose spatial feature maps (H x W x C) for attention-based decoding.Votes: 0GitHub stars: 61
- Cnn Metadata Fusion HeadFuse CNN image features with a small tabular MLP branch via concat before a final classifier, training both branches end-to-endVotes: 0GitHub stars: 61
- Cnn Rnn Video ClassificationExtract per-frame CNN features then classify the temporal sequence with stacked GRU layers and a boolean mask for variable-length video inputsVotes: 0GitHub stars: 61
- Coco Annotation ConversionConvert per-instance RLE or polygon annotations to COCO JSON format for seamless use with Detectron2 and MMDetectionVotes: 0GitHub stars: 61
- Combinatorial Deletion Point MatchingMatch two unequal-length point sets by enumerating which points to drop from the longer set, searching over deletion combinationsVotes: 0GitHub stars: 61
- Concat Tile Feature PoolingPasses N tiles independently through a shared CNN backbone, concatenates their feature maps spatially, then pools for classification — a lightweight multi-instance learning approach.Votes: 0GitHub stars: 61
- Confidence Threshold MultilabelAssigns multiple labels per sample using a confidence threshold on sigmoid outputs with a fallback negative class.Votes: 0GitHub stars: 61
- Connected Components Centroid ExtractionExtracts object centers from 3D segmentation masks using connected-component labeling and centroid computation.Votes: 0GitHub stars: 61
- Conv1d Lstm Stroke ClassifierStack 1D convolutions for local feature extraction before bidirectional LSTMs to classify variable-length stroke sequences into hundreds of doodle categoriesVotes: 0GitHub stars: 61
- Coverage Stratified SplitStratify train/validation split by binned mask coverage percentage to ensure balanced foreground representation in segmentation tasksVotes: 0GitHub stars: 61
- Cross View Count Consistency FilterDrop false-positive detections by requiring matched per-frame box counts across paired synchronized camera views (Endzone vs Sideline) — if both views see the same scene at the same instant, true events should appear in bothVotes: 0GitHub stars: 61
- Ct Z Stack Orientation FlipDetect inverted CT slice ordering by comparing ImagePositionPatient[2] (the Z coordinate) of the first and last DICOM slice in a series, flipping the volume along axis 0 when needed so every patient ends up in canonical head→feet orderVotes: 0GitHub stars: 61
- Cumulative Sum ChannelAdd a cumulative sum channel along the vertical axis to capture directional structural trends in grayscale images for segmentationVotes: 0GitHub stars: 61
- Dbscan Ensemble FusionFuses 3D object detections from multiple models by clustering nearby predictions with DBSCAN and taking cluster centroids.Votes: 0GitHub stars: 61
- Deepsort Majority Vote RelabelRun DeepSort on per-frame detections then overwrite each track cluster's label with the most common mapped label across the track's lifetimeVotes: 0GitHub stars: 61
- Detectron2 Custom Data MapperCustom Detectron2 data mapper with photometric augmentations that properly transforms images, bounding boxes, and instance masks in syncVotes: 0GitHub stars: 61
- Deterministic Hash PartitioningPartition a large dataset into N balanced shards using integer key modulo arithmetic for reproducible, class-interleaved splits across CSV filesVotes: 0GitHub stars: 61
- Dice LossDice coefficient loss for pixel-level segmentation that directly optimizes the overlap between predicted and ground-truth masks.Votes: 0GitHub stars: 61
- Dicom 12bit Pixel CorrectionFix pixel overflow artifacts in 12-bit unsigned DICOM files where values wrap around at 4096Votes: 0GitHub stars: 61
- Dicom Freq Histogram NormalizationNormalize DICOM pixel values using frequency-equalized histogram bins for globally consistent non-linear intensity mappingVotes: 0GitHub stars: 61
- Dicom Hounsfield NormalizationConvert raw DICOM pixel arrays to Hounsfield Units using per-slice RescaleSlope/RescaleIntercept, with outside-scanner clampingVotes: 0GitHub stars: 61
- Dicom Ipp Z Slice SortOrder DICOM slices into a coherent 3D volume by sorting on ImagePositionPatient[2] (the Z coordinate in patient space), with a filename-integer fallback for series whose tag is missing — never trust filename alphabetical order, never trust InstanceNumberVotes: 0GitHub stars: 61