get first frame

vision_agent/tools/tools.py

@@ -200,14 +200,15 @@ def owl_v2_image(
  )
  data = data_obj.model_dump(by_alias=True)
  detections = send_inference_request(data, "tools", v2=False)
- detections = detections["<CAPTION_TO_PHRASE_GROUNDING>"]
+ # get the first frame detections
+ detection = detections[0]
  bboxes_formatted = [
  ODResponseData(
- label=detections["labels"][i],
- bbox=normalize_bbox(detections["bboxes"][i], image_size),
+ label=detection["labels"][i],
+ bbox=normalize_bbox(detection["bboxes"][i], image_size),
  score=1.0,
  )
- for i in range(len(detections["bboxes"]))
+ for i in range(len(detection["bboxes"]))
  ]
  return [bbox.model_dump() for bbox in bboxes_formatted]
 
@@ -428,15 +429,16 @@ def florence2_sam2_image(
  )
  req_data = req_data_obj.model_dump(by_alias=True)
  detections_ft = send_inference_request(req_data, "tools", v2=False)
- detections_ft = detections_ft["<CAPTION_TO_PHRASE_GROUNDING>"]
+ # get the first frame detections
+ detection = detections_ft[0]
  return_data = []
- all_masks = np.array(detections_ft["masks"])
- for i in range(len(detections_ft["bboxes"])):
+ all_masks = np.array(detection["masks"])
+ for i in range(len(detection["bboxes"])):
  return_data.append(
  {
  "score": 1.0,
- "label": detections_ft["labels"][i],
- "bbox": detections_ft["bboxes"][i],
+ "label": detection["labels"][i],
+ "bbox": detection["bboxes"][i],
  "mask": all_masks[i, :, :].astype(np.uint8),
  }
  )
@@ -1187,6 +1189,8 @@ def florence2_phrase_grounding(
  v2=False,
  metadata_payload={"function_name": "florence2_phrase_grounding"},
  )
+ # get the first frame detections
+ detection = detections[0]
  else:
  data = {
  "image": image_b64,
@@ -1195,14 +1199,14 @@ def florence2_phrase_grounding(
  "function_name": "florence2_phrase_grounding",
  }
  detections = send_inference_request(data, "florence2", v2=True)
+ detection = detections["<CAPTION_TO_PHRASE_GROUNDING>"]
 
- detections = detections["<CAPTION_TO_PHRASE_GROUNDING>"]
  return_data = []
- for i in range(len(detections["bboxes"])):
+ for i in range(len(detection["bboxes"])):
  return_data.append(
  ODResponseData(
- label=detections["labels"][i],
- bbox=normalize_bbox(detections["bboxes"][i], image_size),
+ label=detection["labels"][i],
+ bbox=normalize_bbox(detection["bboxes"][i], image_size),
  score=1.0,
  )
  )