15 रिपॉजिटरी
Standardized evaluation suites for measuring the accuracy and generalization of visual recognition systems.
Distinguishing note: Specifically targets vision-based model evaluation rather than general-purpose ML benchmarking.
Explore 15 awesome GitHub repositories matching artificial intelligence & ml · Computer Vision Benchmarks. Refine with filters or upvote what's useful.
CLIP is a neural network architecture designed to map visual and textual data into a shared latent vector space. By utilizing transformer-based feature extraction and multi-modal tokenization, the system aligns images and natural language strings, enabling cross-modal similarity analysis and semantic classification. The project functions as a zero-shot classification engine, identifying image content by calculating the cosine similarity between visual features and arbitrary text labels without requiring task-specific retraining. Beyond inference, it serves as a research toolkit for evaluating
Evaluating how well visual recognition systems generalize across diverse datasets and identifying performance gaps in real-world application scenarios.
This project is a comprehensive, community-driven repository that serves as a centralized catalog for computer vision research and development. It functions as a structured index of academic papers, open-source software libraries, public datasets, and educational tutorials, providing a navigation point for the complex landscape of modern vision technology. The repository distinguishes itself through a taxonomy-based indexing system that maps the relationships between foundational research, influential academic figures, and their corresponding software implementations. By utilizing a lightweig
Acts as a comprehensive research catalog for influential figures, algorithms, and benchmarking suites.
This project is a computer vision benchmark and image classification dataset used to measure and compare the accuracy of machine learning models. It provides a standardized collection of labeled fashion product images and training data formatted to be compatible with the MNIST dataset structure. The dataset consists of fixed-dimension grayscale images and label-based category mappings, stored in a binary format. It includes pre-split training and testing sets and a static distribution to ensure consistent cross-model benchmarking. The repository supports image classification benchmarking and
Serves as a reference dataset to measure and compare the accuracy of image classifiers.
BasicSR is a PyTorch-based image restoration toolbox and framework designed for training and deploying deep learning models to upscale, denoise, and deblur images and videos. It serves as a comprehensive system for image super-resolution and video quality restoration, providing the necessary infrastructure to recover fine visual details and increase pixel density. The project distinguishes itself through specialized toolkits for facial image enhancement and high-fidelity face synthesis, as well as a dedicated video quality restoration suite that utilizes deformable convolutions and generative
Computes standard restoration benchmarks including PSNR, SSIM, LPIPS, NIQE, and FID.
Boxmot is a multi-object tracking framework designed to follow multiple objects across video frames using motion and appearance algorithms to maintain consistent identities. It functions as a system for tracking objects with specific orientations using rotated bounding boxes and corresponding intersection-over-union computations. The project includes a re-identification model optimizer that converts neural networks into formats for hardware-accelerated execution. It also features an evolutionary hyperparameter tuner that iteratively mutates tracker settings to maximize accuracy for specific d
Uses standardized evaluation suites to measure the accuracy and consistency of visual tracking systems.
This project is an object detection evaluation library and benchmarking tool designed to calculate precision, recall, and average precision for computer vision models. It provides a suite of utilities for parsing bounding box coordinates from text files and calculating spatial overlap to determine detection accuracy. The toolkit features a command line interface for comparing ground truth files against model predictions. It includes a precision-recall curve generator to visualize the relationship between precision and recall across different confidence thresholds and an intersection over unio
Provides a standardized evaluation suite for measuring the accuracy and generalization of object detection models.
यह प्रोजेक्ट एक PyTorch पर्सन री-आइडेंटिफिकेशन फ्रेमवर्क है जिसे विभिन्न कैमरा व्यूज में व्यक्तियों की पहचान करने वाले मॉडल्स को प्रशिक्षित और मूल्यांकन करने के लिए डिज़ाइन किया गया है। यह एक पूर्ण मॉडल ट्रेनिंग पाइपलाइन, छवियों को संख्यात्मक वैक्टर में बदलने के लिए एक डीप लर्निंग फीचर एक्सट्रैक्टर और पहचान पुनर्प्राप्ति सटीकता को मापने के लिए कंप्यूटर विज़न बेंचमार्किंग टूल्स का एक सुइट प्रदान करता है। फ्रेमवर्क में एक विशेष ट्रांसफर लर्निंग टूलकिट शामिल है जो प्रीट्रेन्ड मॉडल्स को फाइन-ट्यून करने के लिए लेयर फ्रीज़िंग, स्टेज्ड लर्निंग रेट ऑप्टिमाइज़ेशन और डिफरेंशियल लर्निंग रेट्स का समर्थन करता है। यह एक एक्स्टेंसिबल इंजन के माध्यम से खुद को अलग करता है जो कस्टम ट्रेनिंग लॉजिक के विकास और हार्ड-सैंपल ट्रिपलेट लॉस माइनिंग और लेबल स्मूथिंग जैसे विशिष्ट ऑप्टिमाइज़ेशन उद्देश्यों के इम्प्लीमेंटेशन की अनुमति देता है। सिस्टम व्यापक डेटासेट प्रबंधन को कवर करता है, जिसमें स्टैंडर्ड बेंचमार्क, बैलेंस्ड बैच सैंपलिंग और इमेज ऑगमेंटेशन के लिए समर्थन शामिल है। यह पुनर्प्राप्ति रैंक और फीचर दूरियों की गणना के लिए मूल्यांकन उपयोगिताएं, साथ ही एक्टिवेशन हीटमैप और रैंक की गई पुनर्प्राप्ति गैलरी उत्पन्न करने के लिए विज़ुअलाइज़ेशन टूल्स प्रदान करता है। यह प्रोजेक्ट Python में इम्प्लीमेंट किया गया है और अपने डीप लर्निंग ऑपरेशंस के लिए PyTorch का लाभ उठाता है।
Provides a suite for evaluating identity retrieval accuracy using standard re-identification benchmarks.
This project is a computer vision dataset and image annotation repository designed for training and evaluating machine learning models. It provides a large collection of labeled images, serving as an object detection benchmark and a source of pixel-level segmentation data. The repository distinguishes itself as a multimodal visual dataset by pairing images with synchronized voice, text, and mouse traces to support narrative understanding. It further enables the analysis of model fairness through the inclusion of demographic attributes and exhaustive annotations. The dataset covers a broad ra
Serves as a standardized benchmark for computing precision and recall in object detection and classification models.
DeiT एक PyTorch विज़न ट्रांसफार्मर फ्रेमवर्क है जिसे छवि वर्गीकरण के लिए डिज़ाइन किया गया है। यह एक ट्रांसफार्मर-आधारित आर्किटेक्चर को लागू करता है जो कनवल्शनल फ़िल्टर के बजाय सेल्फ-अटेंशन लेयर्स और स्थिति-जागरूक अनुक्रम मॉडलिंग का उपयोग करके छवियों को चपटे पैच (flattened patches) के अनुक्रम के रूप में संसाधित करता है। यह प्रोजेक्ट नॉलेज डिस्टिलेशन फ्रेमवर्क के माध्यम से डेटा-कुशल प्रशिक्षण पर केंद्रित है। यह सिस्टम एक छात्र मॉडल को उच्च-प्रदर्शन शिक्षक मॉडल के सॉफ्ट लेबल की नकल करने की अनुमति देता है ताकि सटीकता और सामान्यीकरण में सुधार हो सके, विशेष रूप से छोटे डेटासेट पर प्रशिक्षण देते समय। लाइब्रेरी छवि वर्गीकरण प्रशिक्षण, क्रॉस-एंट्रॉपी लॉस ऑप्टिमाइज़ेशन और इन्फरेंस के लिए प्रीट्रेन्ड वेट्स की तैनाती सहित पूर्ण विकास लाइफसाइकिल को कवर करती है। इसमें मानक डेटासेट के खिलाफ मॉडल प्रदर्शन और सटीकता का मूल्यांकन करने के लिए एक बेंचमार्किंग टूल भी शामिल है।
Includes tools for evaluating model accuracy against standard computer vision benchmarking datasets.
PerceptualSimilarity is a deep learning framework designed to quantify and evaluate the perceptual distance between images. It provides a system for measuring how similar two images or image patches appear to human vision by using deep feature representations instead of pixel-wise differences. The project implements a differentiable distance metric that functions as a loss function, allowing image pixels to be optimized via backpropagation to reach a target visual appearance. It includes a trainable linear layer that can be applied to frozen deep features to learn weighted distance metrics al
Tests the accuracy of visual similarity models against standardized human judgment datasets.
RAFT एक PyTorch कंप्यूटर विज़न फ्रेमवर्क और डीप लर्निंग सिस्टम है जिसे ऑप्टिकल फ्लो अनुमान के लिए डिज़ाइन किया गया है। यह एक GPU-त्वरित गति अनुमानक के रूप में कार्य करता है जो वस्तु की गति निर्धारित करने के लिए वीडियो फ़्रेम के बीच प्रति-पिक्सेल गति वैक्टर की गणना करता है। यह कार्यान्वयन उच्च-आयामी सहसंबंध गणनाओं से जुड़े मेमोरी और कंप्यूट ओवरहेड को अनुकूलित करने के लिए आवर्तक ऑल-पेयर्स फ़ील्ड ट्रांसफ़ॉर्म और कस्टम CUDA कर्नेल का उपयोग करता है। यह हार्डवेयर-स्तरीय त्वरण फॉरवर्ड पास के दौरान GPU मेमोरी उपयोग को कम करता है। यह टूलकिट सुपरवाइज्ड फ्लो लर्निंग और मिश्रित-सटीक प्रारूपों का उपयोग करके मॉडल प्रशिक्षण को कवर करती है। इसमें वीडियो गति विश्लेषण और मानक ऑप्टिकल फ्लो डेटासेट के विरुद्ध मॉडल सटीकता को बेंचमार्क करने की क्षमताएं भी शामिल हैं।
Evaluates the accuracy of motion estimation models against standardized computer vision datasets.
Vim is a state space model vision framework designed for image classification and visual representation learning. It functions as a computer vision research tool that converts two-dimensional image grids into one-dimensional sequences to extract spatial features. The system implements a linear-scaling image classifier that replaces quadratic attention mechanisms with state space operations. This approach utilizes bidirectional sequence modeling and selective gating mechanisms to process visual data. The framework covers computer vision benchmarking and image classification research, providin
Measures vision model accuracy and performance against standard industry datasets.
mmtracking is a PyTorch video perception framework designed for training and deploying computer vision models that analyze sequential image data. It provides specialized tools for multi-object tracking, video instance segmentation, and a configuration-driven system for managing deep learning models. The project utilizes a deep learning model registry and a configuration-driven pipeline to swap model backbones and detectors without modifying the core codebase. This modular approach allows for the development of custom perception architectures by combining various components and configurations.
Provides a standardized evaluation suite for measuring the tracking precision of visual recognition systems.
This project is a multi-object tracking library and computer vision toolkit designed to maintain consistent identity IDs for objects across video frames. It provides a motion-based object tracking system that converts raw detections into stable temporal tracks, enabling the analysis of object movement and behavior over time. The toolkit distinguishes itself through advanced identity maintenance, utilizing Kalman filters for linear motion tracking and sparse optical flow for camera motion estimation. It features multi-stage object association to recover occluded objects and non-linear motion t
Evaluates the accuracy and precision of tracking algorithms against ground-truth datasets using standardized metrics.
This project is a web-based platform designed for benchmarking, visualizing, and evaluating computer vision algorithms focused on image feature extraction and matching. It provides a unified interface to compare the performance and accuracy of different models by processing image pairs or live video streams. The system distinguishes itself through a modular architecture that allows users to define custom processing pipelines and register external algorithms via configuration files. It incorporates geometric verification techniques to refine visual data and improve the precision of detected co
Provides a platform for comparing the accuracy and performance of various feature extraction and matching algorithms.