How boudiafa/agriscope works
AgriScope is documented as a multimodal agricultural-image understanding model, paired with the AgriGround dataset and its annotation pipeline. The main inference path combines biological and spatial visual features with a prompt, then grounds generated text through mask decoding. The README documents the model architecture and four annotation stages, but the repository tree contains no implementation source; these nodes therefore point to the README, and the graph reflects documented design rather than verified runtime calls.
Main entry points: README.md AgriGround Dataset
Components 17
AgriScope Model
- Biological Encoder
README.md - Global Encoder
README.md - Multimodal Language Model
README.md - Grounding Projection
README.md - SAM2 Mask Decoder
README.md - Grounded Outputs
README.md
Annotation Pipeline
- Caption and Metadata
README.md - Phrase Grounding Prep
README.md - Mask Alignment
README.md - Instruction Synthesis
README.md
Dataset and Tasks
- AgriGround Dataset
README.md - Captioning Tasks
README.md - Segmentation Tasks
README.md - Detection Tasks
README.md - Conversation and QA
README.md
Other components
- User
- Hugging Face
Connections 17
- User → Multimodal Language Model: provides image and prompt
README.md - Biological Encoder → Multimodal Language Model: supplies semantics
README.md - Global Encoder → Multimodal Language Model: supplies spatial features
README.md - Multimodal Language Model → Grounding Projection: emits SEG states
README.md - Grounding Projection → SAM2 Mask Decoder: conditions masks
README.md - Multimodal Language Model → Grounded Outputs: generates text
README.md - SAM2 Mask Decoder → Grounded Outputs: produces masks
README.md - Caption and Metadata → Phrase Grounding Prep: feeds captions
README.md - Phrase Grounding Prep → Mask Alignment: prepares prompts
README.md - Mask Alignment → Instruction Synthesis: supplies aligned masks
README.md - Instruction Synthesis → AgriGround Dataset: synthesizes records
README.md - AgriGround Dataset → Multimodal Language Model: trains model
README.md - AgriGround Dataset → Captioning Tasks: covers
README.md - AgriGround Dataset → Segmentation Tasks: covers
README.md - AgriGround Dataset → Detection Tasks: covers
README.md - AgriGround Dataset → Conversation and QA: covers
README.md - AgriGround Dataset → Hugging Face: annotations hosted on
README.md