Theme-first browsing
Start from grouped research families instead of diving straight into a flat list of links.
Daily arXiv Watch
Paper Arxiv Daily
A curated research surface for following fast-moving arXiv output across computer vision, multimodal systems, large language models, and adjacent machine learning tracks.
Timeline awareness
Use time as a design layer, not just metadata, so topic momentum is easier to perceive.
Dense but calm tables
Long reading surfaces stay usable because the heavy information is visually tiered.
Research Timeline
The site works best when it behaves like a visual research timeline: broad archive coverage, live updates, and clear routes from scanning to deep reading.
Foundation Layer
Early archive coverage establishes the baseline for long-running topics like classification, segmentation, depth, and tracking.
Topic Expansion
Multimodal, generation, LLM, audio, and generalization tracks become easier to compare as the archive density increases.
Live Research Radar
The latest stream becomes a near-live reading queue, with analytics and archive pages helping you spot momentum and topic shifts.
Core Entry Points
Use these routes when you want the site to feel like a compact research product rather than a raw markdown dump.
Main Paper Index
The complete topic-organized list with daily paper tables, archive navigation, and improved mobile readability.
Analytics Dashboard
Review trend curves, topic rankings, approximate code coverage, and top authors using the interactive dashboard.
Repository Workflow
Install dependencies, rerun data generation locally, and adjust tracked keywords through the project configuration.
Theme Constellations
Topics are grouped into visual families so the site feels more like a research atlas and less like one undifferentiated list of links.
Vision Systems
Dense, high-frequency topics for tracking visual representation learning and scene parsing.
Dynamics and Geometry
Follow movement, depth, pose, and long-horizon understanding across video and embodied settings.
Generative and Multimodal
Track the shift from single-modality pipelines toward general-purpose generation, reasoning, and modality fusion.
Adaptation and Decision
Use these tracks to follow transfer, reasoning structure, policy learning, and graph-based learning systems.
How To Use It
The site is strongest when it works as a layered reading flow: choose a family, enter a topic, then drop into the archive or analytics.
Pick a family
Use the constellation blocks to decide which part of the research landscape you want to inspect first.
Open a topic
Move into a focused topic page where monthly archives provide a more manageable browsing cadence.
Read or analyze
Choose between dense raw tables for immediate reading or analytics for trend and author-level pattern spotting.
Usage Snapshot
For local regeneration or customization, the existing workflow remains simple and script-first.
Generate locally
Install dependencies, run the fetcher, and adjust tracked keywords in config.yaml when needed.
pip install -r requirements.txt
python get_paper.py
python scripts/count_range.py 2024-01-01 2024-12-31